Karol Dziedzic: Google Ads for e-commerce – Performance Max without assets, ROAS and choosing a specialist

Episode recorded in Polish, published 23 September 2025; the Google Ads and Google Analytics advice reflects the state at the time of recording, and the teaser at the start of the video is not repeated here.

Who is Karol Dziedzic?

Tomek: Hi, and welcome to another episode on our channel, Flying with Shopify. Today we've prepared a topic about paid campaigns in Google Ads and, inseparably connected with them, Google Analytics, which I think sends a chill down the spine of every e-commerce owner. How should this tool be set up properly, and how do you read the data? Our guest today is Karol Dziedzic, a man of many talents, but an expert in these two areas. In your bio, Karol, I found that you're a trainer, a specialist, a lecturer, a consultant and a speaker. Tell me, which of these roles best reflects your skills, which one do you identify with most?

Karol: I didn't list my main skill there, which is being a “clicker”, because that's my favourite thing to do. Just opening the tables in peace and quiet and diving into them to, as people say, move the sliders around. But I enjoy the other things too. If we break all these activities down, the training part is great, because you get to talk to another person and help their business. You don't get that sitting alone at the computer. On the other hand, people are people, there are ups and downs when it comes to the emotions involved, so you can't run training sessions all the time, and sometimes you need to do some clicking.

You also have to click so you're not a so-called training hypocrite. It's worth knowing what you're talking about once you've clicked through and tested it yourself. Technically, being a specialist means being a clicker. Being a lecturer, a trainer, an instructor simply means passing knowledge on so other businesses can grow. In short: I click to check what works, and then I share that knowledge with others so they can benefit from it too. It's a bit of a win-win: I invest some money to see whether something works and pass it on to someone, of course in exchange for some kind of payment. When the other person uses that knowledge, they multiply that financial surplus. And that's how we simply keep pushing the e-commerce world forward.

Why do you have to stay constantly up to date in Google Ads?

Tomek: So you've kind of found a workaround for that famous saying that if someone doesn't know anything about a subject, they should become a lecturer in it, right?

Karol: That's hard in this industry, because if you look at Google Ads alone, there are quite a lot of changes. And we're not talking about two changes a year, but on average about one small thing a day: something moved in the interface, some new feature and so on. I always smile that if someone in our industry goes on too long a holiday, say two weeks, coming back can be interesting. In the meantime a new feature might launch that turns out to be so crucial that everything we taught, lectured on and recommended has to be changed. That's another interesting thing: in 2024 you could recommend certain solutions to businesses, and in 2025 the same recommendation will destroy them. So not being up to date can simply be dangerous.

The biggest thing I'm afraid of as a lecturer is that I'll be behind on something and someone in the lecture hall will call me out on it. Someone will ask me: “Have you heard about this new feature?” Let's say the lectures are at the weekend, the feature launched on Friday, we meet on Sunday and someone asks: “Have you tested it? Do you know what this feature that launched on Friday is about?” I can smile and say that's my know-how, or that I'll check and get back to them. But I'd rather avoid that situation and not be surprised that the feature even exists. In Google Ads, for better or worse, ignorance doesn't get far: either someone catches it, or someone simply loses money because we didn't know something had changed. So the difficulty is staying up to date over the long term, and that's the challenge, because we're not relearning everything from scratch, but very often we have to change our minds.

Tomek: Can you give an example of a big change that came as a surprise and meant that yesterday's recommendations turned out to be outdated after a holiday?

Karol: Over time you could certainly pull out plenty of such things. You also have to consider whose recommendations they are: Google's or Google representatives', or some specialists'. We always say: one specialist says this, another says that, and in the end the client pays to find out who was right. It's not exactly a great situation, but it often happens. When it comes to recommendations in 2025, meaning what everyone is focusing on intensely, the first thing is everything that has AI in it and everything that has the word “max” in it. When those two words come up, everyone always has a different opinion.

The top of the top in 2025, where I see e-commerce businesses' money invested in an extremely inappropriate way, is the classic Performance Max campaign. Last year you could hear that most people recommended it. And here I'll use a bit of jargon, because we're talking about features in the system, but it's worth passing this phrase on to someone, repeating it and asking about it: do we have assets added to our Performance Max campaigns or not? It's a nice, light question, but if someone recommended it to us last year, let's take that as an example, this year it may be destroying our business. If we're working on ever lower margins and have ever higher costs, that one click can completely ruin us. It's just one thing, but the way certain features work has changed: in the past it may not have been harmful or may not have affected anything, and now it can simply be ruining us.

For services too. Stepping outside online stores for a second: there we also have the Performance Max campaign, and the instructions there are simpler. For most service businesses it's simply inefficient from a business point of view. On the other hand, we'll get a recommendation from 2024: do it before others to make use of the time before everyone else adapts. Stuck in the middle, it's very hard for the client to decide whom to believe. Google's recommendation and Google's representatives, because it comes from the source, or the specialists? And which specialist, since each one says something different, one communicates this way, another that way, one is more convincing, another less so? At that early stage we can't verify who's right. That's why this world is a bit fluid, or difficult, and it's easy to make bad decisions, even though, based on the data we had, they seemed right at the start.

Can a store owner run Google Ads on their own?

Tomek: Karol, your answer has already touched on something I wanted to ask, but I'll try anyway, because I think it can be answered more broadly. You're talking about the very frequent changes that Google really does introduce. Sometimes they're bigger, sometimes cosmetic, but they definitely require you to stay very up to date, because they directly affect our business. We often run into a situation where Google Ads is perceived as a relatively easy tool to use. That's also Google's own rhetoric to encourage people: it's so simple that as a store owner I can do it myself, or I should be able to handle it.

You've partly answered that already by talking about how many changes there are, how often they happen and that you need to keep up. That seems to be, or may be, at odds with how store owners approach it. I assume they've gained the knowledge of how to handle this tool, but they rather don't do it on an ongoing basis, and often not at a large scale. So the first, clarifying question: do you recommend that store owners start by doing it themselves, putting in the effort and learning the tool? Because from their business's point of view it's a fairly key tool for acquiring traffic. Or is it worth looking for a specialist straight away? I'll probably ask in a moment how to go about it, but should they look for someone who works with this every day, because by definition that will be more efficient?

Karol: I'll say something contrarian, as someone who runs these campaigns: it's absolutely not always worth hiring a specialist. I paused on purpose so that would sink in. It's absolutely not necessary to have a contractor, an agency, a freelancer, someone to take care of it. In some cases we can run these campaigns correctly ourselves. First of all, it all comes down to the product we sell, because let's be honest: a great product at a great price will sell great even if we screw up the campaign. We'll just make less on it. There's no great magic to it. Even if we set up the clicks badly and run the campaign any old how, but we have no competition or we have a business advantage, it will still sell by the nature of things, the profit will just be lower.

In practice, when I analyse the various enquiries that come to me, I very often leave a recommendation that it's better if I don't do it. That's not super well thought through from a business perspective, if you look at it: a client comes to you and you say you won't take them on. It's a bit contrary. But I'll explain what it's about. It's simply about scale. If we're talking about businesses, let's say online stores, the level of sophistication grows with the scale of our campaigns. That means both the budget and the number of units sold that our activities generate.

Let's say we're launching a new e-commerce business, and let's make this split: we have a budget of PLN 1,000–5,000 to invest. Let that be the entry threshold. With budgets like that, we realistically won't need a specialist to run the campaigns. What may be useful is the starting element, meaning someone to set up our analytics and the initial technical elements in the account. But running the campaigns at such a small scale comes down to clicking on basically two things: which products we want to advertise and which search terms we want to show them for. You can learn to run a campaign like that in an hour or two. It's not exceptionally difficult, it's a simple Excel sheet. I'm leaving out implementation, of course, because that can be outsourced as a one-off. What's more, many platforms have this automated: you just log in, connect Analytics, as people say, enter its ID in the extension widget's field, the same with Google Ads, and it should work.

It's basically about costs. If we look at the cost of managing campaigns up to, say, PLN 5,000, the industry average is roughly 10 to 20% of that amount. Arrangements vary: subscription, a share of profits and so on. If we invest PLN 5,000 and on top of that we have to pay PLN 1,000, sometimes PLN 1,500 or 2,000 for management, that's already a big difference in how much less we invest in promoting our products. If we replace that with a one-off implementation cost plus a consultation from time to time, we can invest that surplus in growing the business and reach higher revenue faster.

Besides, “clicking” in these campaigns isn't a full-time job. Absolutely not. Maybe I'm shooting myself in the foot, but since we're talking so transparently, let's be honest from cover to cover. A campaign that's already running, most often Performance Max for an online store with a budget of up to PLN 5,000, can be managed fully and well by spending one to two hours a week on it at most. It's not 10 or 20 hours. It's literally an hour on Monday, an hour on Friday, and we're on top of everything, as long as we do it regularly. What's more, as business owners we have an advantage over absolutely every contractor: we know the product and the audience better, we can play with price and quality, we're more up to date on the competition. We're the ones who can better verify what to sell and at what price. It's actually hard to be a better contractor than a client who, I'll stress, has the time and the will to do it. If someone has absolutely no wish to learn this, they should have a contractor, because otherwise it may end up being inefficient.

An agency, a freelancer or someone in-house?

Karol: Later, when the business grows and we already have budgets in the PLN 10,000–50,000 range, there start to be a few more challenges. First, more campaign types come in and their number grows. You have to start tying it more closely to business matters: not just how much cash to transfer to Google, but how it then adds up in terms of returns, margins and business costs. Worse still, if we add more marketing channels, the classic case: we've grown Google and we're adding Facebook, the question arises of what to invest more in and what less in a given period. So we have more analytical elements to check. And here you have to ask yourself how much passion, will and time you have as a business owner to handle it.

We usually have three ideas. The first: let's outsource it to an agency. There's absolutely nothing wrong with that, we simply hand over the whole management externally. Everything has its pros and cons, but it's a “just make it work” kind of solution. Number two is handing it to a freelancer, a sort of external employee of ours, with whom we usually have somewhat closer contact. Pros, cons: if the freelancer goes on holiday, usually nobody directly replaces them, which also leads to interesting situations. It's neither better nor worse than the agency option, it's just different. The third option is doing it on our own or in-house, meaning training someone or finding someone with free time to dig into it.

None of these solutions is ideal. It all comes down to costs, resources, and the speed and efficiency of communication. You have to pay close attention to these three elements, because each of the solutions can absolutely end well, meaning we'll make money either way. The question is what happens in between: whether we'll be frustrated or dissatisfied, and whether everything will go the way we, hopefully, planned before making the decision.

So, scale. If we jump above that 50,000 into budgets like 100,000, 200,000 and more, then there really is something to play with every day. And again the question is who should do it and what costs that brings, because growing ad budgets in fact also increase management costs. We reach the point where we start wondering whether it's better to delegate it or, at that amount, better to have someone in-house. You could say that in most cases it's better to have someone in-house, but the question is whether there's anyone on the market who'd want to work with us that closely and for that much time. So there's always some question. I smile that ideally the company boss would fall in love with Google Ads, drop all their other duties and do all the clicking themselves, but that's absolutely never the solution. Unless, like me, you're a one-person business that only does Google Ads. Then I'll find the time to serve myself, it's not that hard.

When asking about the ideal setup, you need to check a few parameters, and we'll always answer: it depends. Just like with a store platform. When someone asks: “Which platform do you recommend, which one is the best?”, nobody will say: “Take the first one or the second one.” First comes the question: “What criteria and parameters does the solution need to meet?” On that basis we can choose something that fits. So choosing how to run your campaigns is literally like choosing an e-commerce platform: it depends on your needs.

What is the minimum an e-commerce owner should know about Google Ads?

Maciek: I'd like to dig a little deeper into this. In that ideal scenario where the owner runs Google Ads themselves, what do you think is the minimum knowledge every e-commerce owner should have when it comes to Google Ads?

Karol: When it comes to Google Ads? Okay. Every owner should definitely know what the most effective ways of promoting their business are today, meaning which campaign type we should go for at a given stage of the business's growth. For e-commerce we're talking about Shopping and Performance Max campaigns, so that's fairly simple. Second: they should know how it works, meaning things like conversion rates, return on investment, LTV, which is customer lifetime value. We should know all of these metrics and how they work in our business, so that we have realistic expectations of the campaigns.

Because without that awareness… Let's take a real-life example: someone runs a brick-and-mortar shop. If 100 people walk into that shop, definitely more than half will walk out with something in their hands. Easily, depending on the shop, of course. In online stores, on the other hand, if one or two out of those 100 people leave with something in their hands, that's a success. If we're coming from offline, that expectation is completely different. Because if only two out of 100 people bought something in a typical shop, and I'm not talking about extremes, it would be puzzling, to say the least, that most people were just looking.

Besides those fundamental expectations, meaning knowing how these metrics add up at the end of the day, we also need information about the potential for advertising our products online. In other words, whether there is any potential for advertising these products at all, and how big it is. That's a matter of doing market research, of course, but without that knowledge it's hard to plan anything. Because if our expectations are very high and the market isn't ready for our products, that also changes the strategy we should pursue. So it comes down to a basic knowledge of the tool, its capabilities, and how our business can be promoted with it. In other words, what can be done with it based on this input data.

But should a business owner know how to wire all of it together technically in analytics? Absolutely not. That's a hobby. They don't have to, it can be outsourced. Some people change the wheels on their car themselves, some have it done. That's literally all there is to it. No one's crown will fall off. The main thing is that in the end it works and it drives. Should they be able to create campaigns and so on? They don't have to either, because someone can do it on their behalf. It's not required.

How do you check what your ad spend actually delivered?

Karol: What they definitely must know at the end of the day, though, is how to verify it. And that's not obvious, because we'll see different data in many places and none of it will be 100% true. What I mean is: we outsource a campaign, we run a campaign, we look at Google Ads and see a great number. We look at Analytics and there's something completely different. We look at the Facebook dashboard: yet another story altogether. So the most important lesson of this starting phase is also figuring out what distortions there are in our business. In other words, which of the numbers is the outlier and how it ultimately translates into our business. It's combining data from several sources, let's say in good old Excel, and checking how it actually translated into our business.

Setting Google Ads itself aside, this is knowledge everyone should have anyway, even if they don't deal with Google Ads. These days the most important thing for business owners is being able to verify the business side. In other words, how much I paid whom and for what, and where it all flowed, meaning how it affected my business. Of course, checking this is fairly intuitive if we run only one advertising channel: we've transferred funds to Google, our store has no other advertising channels, so we can attribute all revenue to Google. That's simple and easy to translate. But, as I mentioned earlier, we add a second channel, let's say Meta. And now the question is: if in August we got such-and-such data, is it better to invest more in Google or in Facebook in September? Which metrics should we focus on, and what should we focus on in a given channel? That's what I currently see as the biggest difficulty.

Here again there are certain pros and cons, if we go back to, let's call them, the types of outsourcing. If we outsource everything to an agency that does both Google and Facebook, then technically the calculations and the advice on what we should invest in will be on their side. So that's off our plate. If we hire a separate freelancer for Google and for Facebook, each will show us different data. Usually, when we add up the result from the Google person and the Facebook person, we get twice as much as we have in the bank, which is puzzling, to say the least. And getting into discussions about who attributed a bit more to themselves also requires a certain fortitude, as I'd put it, because these are very interesting discussions.

And if we try to do it ourselves, it can also be a frustrating process. We go online, read on various websites how to do it, and everyone has a different opinion. We ask ChatGPT, and the result of that query is a bit of a comedy too. The chatbot learns mostly from material on the internet, from Google and so on, and in practice not everything works perfectly, which can also be frustrating. And then we need a person who will put it all together, someone who can actually get a handle on the analytics. If it were simple and calculable, specialists would absolutely not be needed. We'd switch on a campaign, costs and revenue would match to the last złoty, and that would be that. As it is, everything is fluid. Everyone has a different opinion, and we as business owners are the referees in a dispute over who did better. And it's hard to say who had the better result, especially when these activities also overlap. So that's the entertainment and the curse of our times at once. Not even the campaigns themselves, not running them, but verifying what actually happened after the investment.

How do you choose a Google Ads specialist or agency?

Tomek: Absolutely. I like the way you put it, Karol: you call all forms of working with specialists and freelancers outsourcing, and there are different types of that cooperation. Let's say we've already defined how much a store owner should know, and they come to the conclusion that at the scale they've already reached, or will reach soon, they need support: some specialist or an agency. Because there really are both types on the market. You represent the specialists, the freelancers, and on the other side there are agencies, meaning groups of people who handle marketing activities.

If we could focus on that a bit and give our listeners a tip on how to go about finding the right person for their business, because I think that's important too. As you said, these people often have different skills, which can be matched to different areas, different topics, different stores, for instance. And we're looking for someone for our specific brand and our specific need. What tips, what advice would you give, what should we pay attention to in this process? What elements should be included in the offer from such a specialist or agency, in that cooperation model, and what should set off alarm bells for us store owners that maybe this isn't the right person, or that the cooperation doesn't look promising from the start, because the way they want to work with me isn't quite how it should be? If you could lift the veil a little: how do you sell your services, what do you pay attention to, and what do you think has no place at all in this kind of cooperation?

Karol: That's a question that stirs up a hornet's nest, but I'm very happy to talk about it. For me the most important parameter in a cooperation is communication. It doesn't matter whether it's an agency, a freelancer or any other form of cooperation. We have to be able to talk well and exchange information. There are different approaches here. Some clients simply like to call and talk. And the question is whether the contractor can always pick up the phone in real time and pass on information, or whether it will look a bit different.

Let's take a stereotypical situation as an example. I have a problem: I see, for instance, that a given product is running out in the warehouse, and I want to switch it off in the ads. I want to pass this on to, let's call them, the contractor, and I like to call. So I call. Nobody picks up, or someone picks up who doesn't work on my project, and the time starts to drag on enormously. If we can't switch off a product we didn't want to advertise for several hours, a day or a week, that's frustrating both business-wise and personally. And that's exactly the first thing. The question is whether we call, email, communicate via Slack, Discord or some other messenger, or work on tasks in some system. I'd simply start with that: with the question of how we'll communicate and how often.

Because the other way round: if we're the kind of people who'd like a meeting every week, but our business doesn't require analysing the metrics that often because we operate at a small scale, then we can't be frustrated when someone says: “Okay, let's do it once a month, so we have something to talk about.” When looking for a contractor, we shouldn't take offence that someone doesn't want to talk to us as often as we'd like, because we have to bear in mind that some meetings could have been an email, and four meetings could literally have been one Meet call. I point this out because it's not about not talking, it's about talking on the basis of data worth looking into. At a small scale we hold meetings that also waste our time as business owners, time we could have spent on something else.

So those who like to call should find people who do that, whether a freelancer or an agency. It's a matter of communication. Those who don't like calling should have people on the other side who communicate asynchronously and by email. There are simply different forms of cooperation. Are there other types? There are also clients who commission something and then disappear: ghost clients. That's quite an interesting phenomenon too, but let's say it's a kind of third extreme. To sum up: first, the form of contact; second, the frequency of contact; and on top of that, understanding that meeting every day won't make more buttons appear in Google Ads for us to click. Sometimes it goes like this: we hold a meeting, after three hours we define which persona we want to reach, we open the Google Ads dashboard, and those buttons aren't there. Simple as that. We have a great strategy that can't be translated into the tool.

The second thing, which unfortunately we can't verify, though it would be great, is the effectiveness of our prospective business partner's work. We simply can't do that. We can read a case study, we can talk about other campaigns. But we have to bear in mind that if the contractor has signed a lot of so-called NDAs, there are things they can't share. It's not that we approach a contractor and they don't want to show us results because they're no good; often they simply aren't allowed to. A check along the lines of “Show me the wardrobe you built, because I want a similar one” won't work here.

Even if someone specialises in a given industry, it doesn't mean they'll help our business, because we have to bear in mind that ads in themselves don't sell. They only take the products we advertise to a wider audience. It happens that someone decides: “I'll go to a company that specialises in selling garden furniture sets, because I make them too,” and it turns out they did a good job for the neighbour, while my company is on its knees, so they probably don't spend as much time on it. We have to bear in mind that we only and exclusively buy advertising space at the right time and can advise on what works and what doesn't, but in the end it comes down to the product that's supposed to perform here.

Report or conclusions: what should you expect after the first month?

Karol: The third thing to verify, unfortunately for clients only after the first month, is the form in which we'll talk about what happened. I'll set aside whether it's a meeting, a report, an email with results and so on. It's about an honest presentation of what was achieved, what wasn't, where there's room for improvement, and what should be dropped. In e-commerce, even in the simplest form, it would be a report: we invested this much, this much came back, with a certain margin of error, of course. We have to verify things like how much data we have in analytics and how much there is in reality, because there are plenty of gaps in between. We have to check the business metrics, for example how many returns the client had, which is data we don't always see in analytics. We have to talk about margins, business costs, product shipping, staff costs, warehouse costs and so on. And at the end, see whether the investment actually paid off or not.

It's easy to verify, because there are usually two scenarios. Either we get a report, and a report isn't a conclusion drawn from the work done, it's just a nice graphic representation of costs and potential returns. Whether it's a report from Google Ads, from Looker Studio or from Analytics doesn't matter. It's simply a table of raw data. We can call it a report, and we as business owners can ignore that table and fail to appreciate its colour scheme, because we need hard information. We put in this much, what happened? If we put in more, will it get better? Have we already hit the ceiling and need to do something else? These days we're turning a bit from Google Ads specialists into financial specialists in a narrow field, meaning how money can be invested in one channel. And from that you can tell quite clearly whether the cooperation will be more business-oriented or rather about servicing a single channel. That's quite telling. Of course, before starting a cooperation you can ask in what form this communication will take place at the end of the month and what our expectations are.

Let me stress one small thing once again, because it can help with communication. Let's never ask for reports or make them; let's send business conclusions instead, let's share them. It may sound like a synonym, but it's something completely different. Sometimes you have to tell the client straight out: “Even though it's bad business for us as the contractor, at this stage of your company's growth Google Ads isn't the channel for you, because we're simply burning money. Nothing positive is happening here. Let's pause the campaign, freeze our cooperation, and focus on your product range, your store and so on.” As I mentioned earlier, what we actually need is business and financial advice: how this tool can help us grow our business or, the other way round, that it won't help us. That's it.

Because if someone knows how to manage campaigns properly and our business can be advertised, that third element, the conclusions, will show: we managed to calculate it, we have growth. Let's try to develop it further, or the other way round, let's try to maintain it, because few people look at it from that perspective. Sometimes maintaining revenue month after month requires a huge amount of work, in the campaigns among other things, because of increasingly intense activity from competitors. Some clients get frustrated that they pay so much for management and their store's revenue isn't growing. But in reality, if the specialist hadn't taken care of it, revenue wouldn't even have held steady. It would have been steadily shrinking, so the company would have been sinking.

As part of these conclusions, we need that kind of feedback. Technically, for example, results from Google Ads are growing, but so are the costs around them. And now we have to consider, for instance, whether our products should have a price adjustment, if possible. Shouldn't we bundle products into sets to increase the basket value? Shouldn't we promote products that open the door, meaning ones that help us acquire a customer who will then buy from us many more times? Here again we're back to the absolute basics: we need to know the conversion-related metrics, conversion rates, LTV values, business values. Then we need a partner or in-house knowledge of what this system will let us do. And we actually need to keep checking it periodically, no less often than once a month. And the bigger the project, the more we're talking about periods like a week, 10 days. It can of course get even shorter.

The way of communicating. Effectiveness that can't be assessed. And finally, the way of sharing conclusions, not reports. That's how I'd define these three elements, because when you send out requests for proposals to 50 companies, everyone will say they'd be delighted to take on the challenge, everyone will have a nicely tailored offer and so on. At that stage we can't compare that. The only thing we can compare is prices. Some will be higher, others lower, but they result from different things. The cheapest isn't always the worst and the most expensive isn't always the best. It's always somewhere in the middle. Ideally you'd have someone recommended, meaning a friend in business who is happy with their contractor. That's the easiest way to start.

But, as I mentioned, the risk when searching broadly is that on day one we won't know whether it's going in the right or the wrong direction. We'll find out after the first period, after the first month, from the way someone communicates either success or failure to us. And consider this: if someone communicates a failure professionally – I mean failure in the sense that we invested more than came back – and points out what caused it, whether it was a product situation, a market situation and so on, they're still a great contractor. They did what they could to deliver results, but they have no influence on the industry, the product or seasonality. Or they were able to spot what was causing it, to give the business feedback on what it should change to keep selling through this channel. That's also huge value, often overlooked, because emotions come into play: we invested, it didn't pay off, so someone must have done something wrong. And we keep forgetting that it's not the ads person who sells. The ads person promotes a product that has greater or lesser sales potential.

How long does it take to judge whether a campaign works?

Maciek: Exactly. I'm wondering… maybe I'll start differently. How much time do we need to assess the effectiveness of a campaign, of some settings or ideas we're testing?

Karol: That's an interesting way to put it, too, because most often you'll hear that it's one to three months, sometimes longer. Paradoxically, though, in Google Ads we measure time not in days but in amount of data, which is a bit of a reversed concept. So for practically every question of whether it's worth clicking something, using a certain feature or not, we can get an answer after roughly 50 conversions, and for online stores: after 50 sales from a given campaign. Large companies can get those 50 conversions in a few days. For smaller ones it can take one to three months. That's a kind of benchmark. For example: if we point a given campaign at different keywords, different products, will that be good or not? How long do we have to wait for the result? About 50 sales for the system to test it. That's a kind of statistical sweet spot where we can verify it.

It happens that in small projects those 50 sales can't be reached, which isn't a problem either. Then we look at slightly lower-quality data, so-called micro-conversions, but that number 50 is still our benchmark. You can simplify it to a rule: no longer than a month, no sooner than 50 conversions. And that's roughly how we can test certain things, faster or slower. So in our industry there isn't much room for “let's see how it goes over the next six months” or “let's wait a bit longer, it'll work out somehow.” You have to approach it in a fairly binary way. A certain period passes, let's say a month, or we've crossed the 50-conversion threshold, and we see: it works or it doesn't, we fix it or we continue. So it moves fairly quickly.

Of course, it happens that some things take up to three months to calibrate, but that's a matter of scale. If we sell 10 products in three months, but they're, for example, very expensive products with a high margin, it simply works differently. In typical e-commerce, though, absolutely nobody needs more than two months to test whether it's worth continuing certain activities, and it all comes down to that magic number of around 50 conversions. “It depends” is becoming a little outdated, because we usually have data that says “good” or “not good,” rather than “we'll see, we'll check how it goes, it depends” and so on. It's also a signal that it's worth asking what's behind that “it depends.” Is there some parameter that still needs to be filled with more data, or not?

What red flags in a specialist's communication should worry you?

Karol: And speaking of red flags, one more thing worth paying attention to just came to mind: transparency of communication, which is a bit of an extension of the first point. If someone tells us it's worth doing this or that, investing more here or somewhere else, it's worth asking what data that conclusion was based on. That will nicely show whether someone will openly say: “Look, listen, here are these metrics in Analytics, and these in Ads. Here something is under-reported, here we need to shift a few things around, because it isn't perfect,” and tries to explain to us where it comes from, or whether we'll hear: “I won't tell you, it's know-how, it depends” and so on. Those are a kind of red flag worth paying attention to, because at the end of the day most things can be calculated, with some degree of under-reporting. We won't find out to the last złoty how everything worked, only within a certain margin of error, but that margin of error also needs to be discussed. And that willingness to communicate matters.

It's a bit as if an accountant were hiding certain costs from us, which they shouldn't. It should all be transparent. Do we want to be the accountant who does this tedious work every day to calculate everything neatly? That's a question about our hobbies, right? Looking at it from the side, whether you handle invoices or handle clicking, it's a fairly similar type of work. You sit in some spreadsheets and sort it out. But if we ask an accountant why the tax is this amount and not another, we'll get an answer. And when we ask someone why we have these returns from Google Ads and not others, we're asking literally the same thing, just about different spreadsheets. There's no magic here. We don't fly on broomsticks, we don't wave wands. We just have somewhat more broken data than accountants do. But we can still advise on what to invest in and what absolutely has to be dropped.

How should you pay a Google Ads agency or specialist?

Maciek: A very common topic in conversations with agencies and freelancers is the billing model you mentioned. You said it's usually 10–20% of the budget. What's your view here? What's the industry standard, and what do you think should be the standard? Because opinions differ. I've come across the view, for example, that pay should depend on performance. From what you're saying, I get the impression you'll rather disagree with that, because, as you say, it's not the specialist who sells the product. But maybe you'd like to comment on that.

Karol: Sure. There are solutions that are ideal for each of the two sides, but there's no single one that's ideal for both at the same time. I always smile that the ideal for the client would be paying the specialist per hour, or per every single click, which from the contractor's perspective makes no sense at all, because we spend the bulk of our time looking at numbers, not clicking, which is a kind of paradox in itself. From the contractor's side, it would be wonderful to get paid on the client's success, as long as things are going well for them, and not so great to get paid on their downfall, because then we're in the red. But in reality, that's not what we sell. The fact that someone makes a better product and gets a better return on it is absolutely no reason to slap a percentage of their success on top of it.

What's more, success-based percentage deals tend to get a bit pathological, because some data can be inflated. In some campaigns you can click a few things to make them look like they're returning better than they really are. I'm not saying that happens. These are cases that, as always with such stories, someone saw once. Whether it happens more or less often, I'm keeping my fingers crossed that it doesn't happen at all. But the drive for ever-higher revenue can lead to situations where, for example, revenue in Google Ads campaigns keeps growing while the business stands still, because we're just shifting revenue and its attribution between channels. That really shouldn't happen. In the long run, in terms of business growth, it will just end in a big commission.

Averaging all of this out: billing per click is sometimes funny, because some people would end up at zero. There are situations where for a whole month the specialist doesn't click anything at all in the account, because there was no need to, and there's no point breaking things by force. Sometimes there's pressure from the client along the lines of: “So why am I paying you if you didn't click anything? There's a tab in Google Ads called Change history, and we checked there that nothing happened for a month.” Only that isn't nothing. We have to look at it like this: is it worth paying a firefighter a full month's salary if there wasn't a single fire that month? Yes and no. But when there is a fire, we'd want them to react exactly when it's happening.

We work a bit like that too. This fee, often a monthly retainer, means that, first, we analyse data, stay up to date and react in real time when something happens. We're always on hand to respond to certain things or to support matters outside Google Ads with our advice, and so on. But it's not piecework, because after a while most things in the account are done by automation, and our role comes down to advising, keeping watch and pointing out opportunities when they appear. So the most common billing model in the industry, which isn't perfect but is the most balanced, is a fixed fee that includes a number of hours, or operational time dedicated to a given project.

If more involvement is needed, these amounts usually don't go up, because let's be clear: there are months with more work and months with less, and it averages out within the contract we've signed. It's not that every month we click more and more, quite the opposite, but that changes too. And it's fair to both sides: the contractor has a steady income and gets that fee like a firefighter, for keeping watch. Whether they click or not is another story. The business, in turn, has a constant sense of security that someone is keeping an eye on things and making the changes that need to be made.

It's a kind of hybrid situation in which both sides want things to go in the right direction, and it doesn't produce extremes. We don't click by force just so that someone's happy the piecework got done. And the contractor's pay doesn't drop either, so the motivation to work on the project doesn't drop. Because if there was nothing to click in a given month, the next month someone might focus on another project, because this one wouldn't be profitable for them, whether at an agency or as a freelancer.

Again, I'll compare it to accountants. We generally pay them the same amounts month to month, unless the scope of work grows. For example, we open new markets in our campaigns, change the strategy, need more traffic: then we can negotiate. On the other hand, a fixed business cost is easier to manage. If these costs jumped up and down month to month, it would be hard to handle from a business point of view. Say we double the budget and the management fee doubles too, and at the end of the day the numbers may not add up at all. So although it isn't the perfect solution, fixed fees bring benefits and certain nuances for both sides.

But a percentage of revenue is a very slippery topic, and I'd advise stores to be careful. Hourly billing can also be unfavourable for businesses, and even more so for the contractor. And billing per click is, of course, our industry joke. It simply shouldn't happen, because we're more like the people who tighten the right screw at the right moment so our ship sails in the right direction, not people who rebuild that ship from scratch, screw by screw, every day. That's not the job at all. I think the association with a firefighter or a systems administrator reflects our work better. Someone sits in the server room day after day, but doesn't re-plug the cables from scratch, because that makes no sense. But when something blows up and needs replacing, they're right there to do it immediately. Plus, by constantly watching how everything works, they advise what to do, what not to do, what new things are coming in, what to stop using and what we can improve in our business.

Tomek: That's true. We once sat at work in a server room with a systems administrator, joking that he gets paid for watching a progress bar.

Karol: We do exactly the same: we watch our clients' revenue grow. Like in those memes: we sit with popcorn watching the bar go up, not clicking anything at all and laughing that someone is still paying us for it, right? But when that revenue bar stops growing, the phones start ringing or the inbox starts filling up, I keep my fingers crossed that we'll know how to react to get out of that situation. Because that's exactly what we earn money on: business advice, being present and reacting at the right moment, so that at the end of the month the business keeps its revenue.

What ROAS can a store owner expect?

Tomek: Listen, Karol, I have a question that's a bit on the border between evaluating an agency's or specialist's work and the Google Ads channel itself. From your experience, can you share what return on ad spend store owners can expect? The standard, so-called ROAS. In your observation, are there any established standards or levels a client can refer to when evaluating a campaign that generated sales, while the agency or specialist talks about a ROAS of 500, 600, maybe 1,000%? First: are there any such values we can use as a benchmark? And do they differ by business size, or do they rather depend on what product we sell and what its margin is?

Karol: Answer B. We absolutely don't have a table along the lines of: in this or that industry, if someone does a good job for you, the expected ROAS is such and such. That won't work at all, because we have different products, different stores, different customer service and, in the end, different margins. What's more, hitting a ROAS of 500% can make one client grow and kill another, because there are all those costs underneath. I don't look at it that way at all. These, let's call them, expected ROAS figures, LTV, CAC and so on, all the customer acquisition cost parameters, always have to be set individually. If we aim for the same return as our neighbour, it can kill us, because we don't know that neighbour's costs and what it costs them to run their business.

Sure, there are industries where hitting a ROAS of 200% already borders on a miracle. But if we have a margin of around 80–90%, that business will grow. In another industry, if we don't hit a ROAS of 2,000%, we might as well pack up, because the margin on the products is so low that it simply won't work. And then all the fun starts: do we have our own warehouse or not, what are our shipping costs, what does it cost to pay the people who handle all of this. So while campaign results can still be described as “in this industry it's such and such a level”, from a business perspective that's no information at all, because it isn't compared with our costs. If information on total return on investment, so no longer ROAS but the so-called ROI, were publicly available, then we could compare. Small spoiler: it absolutely isn't available anywhere. We don't live in such a transparent reality that we'd have that data anywhere.

As I said, a ROAS of 1,000% is death for one business and profit for another, so chasing a specific number makes no sense at all. What does make sense is what we talked about earlier: talking about exactly this, at least monthly. We check how much we invested, what revenue we generated, what final profit we were left with and how we can scale it later. So I wouldn't stand over the contractor saying: “Okay, this month ROAS was 700%, so next month make it 1,000%.” That's often impossible from the tools' side, because ROAS can often be increased faster by, for example, starting to advertise more expensive products or bundling products, rather than changing something in the campaigns.

Of course, in campaigns we can set the return we expect, but we also have to take the system's capabilities into account. If we go into Google Ads and type in: “ROAS 1,000%, please”, and the system isn't able to achieve it, it may switch our ads off completely. No ads, no revenue at all, so we'll drown. So we can't transfer our dreams into the dashboard; we have to know within what range this system can help us. And that's where we prove ourselves as advisers: we can tell the client that the range the campaign can currently work in is between 500% and 800%. The question is whether we increase revenue at the expense of final profit, or reduce the budget but squeeze more out of it so the profit is higher. That's what we should be talking about, what will happen depending on the settings we put in, and not along the lines of: “Dear client, what's your margin? Okay, 20%, so I'll punch in a ROAS of 500% and it'll somehow work out.”

Then there are things we often don't know about. Let's say we optimise a campaign only from within Google Ads and don't look at other channels. Then we live in a beautiful but unrealistic world, because other channels affect sales too. We don't see returns, costs or all the other things; we just show in the report: “You wanted 500%, you've got 500%”, while you're drowning and nobody knows why. Our role changes over time. Years ago, if someone wanted a certain return on investment, our job was to deliver it and show it in the report for a given channel. Now we step outside our channel and try to look at things as a whole. Okay, we operate in this channel, but let's talk about the returns from the whole venture so we know what share our Google Ads had in it. Can it be increased or not? Do we need to focus on other channels, and how does it all work together?

We also joke that we spend less and less time in the Google Ads dashboard and more and more in Excel, CRMs and databases covering the whole business, checking along the way where Google Ads can still help. So we're evolving a bit. And I'll smile here too, because most people predict the death of Google Ads specialists, who are supposed to be replaced by ChatGPT and other tools. They can replace them to some extent, but our scope of work is evolving too. We're moving away from clicking towards verifying how things work as a whole. Clicking itself may not be all that fascinating, so if some automation takes it over, nobody will be sad. But when it comes to looking at the whole picture of what can be done at a given moment, we won't be replaced that quickly just yet.

How can AI help with Google Ads and analytics?

Maciek: Since you've brought up ChatGPT, AI and the future of Google Ads: what can you recommend? What tools can we already use now to gain an edge with AI in advertising and analytics?

Karol: Honestly, any of the top LLMs is fine for supporting our work. Whether we use ChatGPT, Gemini, Claude or whatever we have at hand, although we mostly move between those three, we'll get support, we just need to look at it in slightly different terms. None of these tools will fully do the work for us, but they can speed up some time-consuming parts. For example, most of our work managing online stores is analysing whether products are profitable or not, so that some products can be switched off from advertising and others kept. You can do it manually, you can do it semi-automatically, or you can periodically feed that data into some AI to do it for us. That speeds up our work massively and also increases the accuracy of what we do.

The second area is keywords: which ones are worth advertising on and which aren't. Again, we can pull the right data out of Google Ads and send the LLM instructions to process it from the right angle. I joke that it's work we used to do every day, only now we have, in quotation marks, an intern whom we instruct how to do it so they process it for us. The benefit is that in many businesses nobody finds time for this, and thanks to this support we'll find or win back that time. So: analysing products and their performance, and analysing keywords and their performance.

Another element is market analysis, which, to put it plainly, means pulling competitors' product prices and analysing peaks, increases, drops, promotions and so on. We can pull that data; it's about processing it so we have it in a cleaner and faster form. Then attempts at analysing seasonality, general market downturns and so on. These are extra research elements we can use to know when, for example, to add more budget, when to ease off, when to gear up for higher sales and when we'll hit a slump. So AI is becoming a bit like our mini research team for repetitive, tedious, time-consuming work that is at the same time super important for the business, because once we get those signals, we can react precisely. For some businesses it's a massive time saver, and for others, for better or worse, it means not needing to hire that whole team and relying on tools instead, so that the business grows better and faster.

Can campaigns run on autopilot without a human?

Maciek: Do you see a scenario in which campaigns, or performance marketing as a whole, run on autopilot without any human involvement?

Karol: Absolutely. I even run such tests myself, checking how much I'm still needed in some businesses. This is where my training side comes in handy, because professionally I basically teach people how to perform repetitive tasks. That's more or less what it comes down to: once you've built a campaign, regularly check this, click that, check this, click that. When you think about it more deeply, and that thought hit me after my first thousand training sessions, first, I keep saying the same thing over and over, and second, if you wrote it down, it's literally learning along the lines of: if this, then that; if this, then that. And that can bring to mind a simple “if” in a programming language. So I had the idea that since it's that repetitive and it's just clicking, maybe I really can be replaced. And that's more or less how it looks.

I joke that a setup is emerging that someone nicely named 10-80-10. Let me explain what's behind these numbers. It's the breakdown of time spent on a project. The first 10% is strategy, connecting the tools, calibrating analytics and so on. Here we need a human to take a look. Then 80% can be taken over by automation, which will analyse the tedious daily work: if something grows, it scales it up; if it drops, it cuts it, and so on. Automation will do it faster, it will be more motivated, it won't take breaks, and in reality this is very repetitive work. Of course, when it isn't sure, it will send us an email or some notification along the lines of: “Take a look at this anomaly and make a decision, dear specialist, and I'll handle the rest.” The last 10% is linking the data with the business, so not just analysis in the ad dashboard, but how it translated into the whole business. So time-wise, you could say we win back 80% of our time.

But we also have to say clearly that a Google Ads specialist isn't the most brilliant and educated person, but the one who can regularly, over and over, deliver a repeatable result. And that comes down to clicking the right place at the right time depending on the parameters, which can already be automated. So if we combine not even AI, but just automation tools, say n8n or Make, or simply play around with the API, then in many businesses that repetitive daily work can be automated, whether in stores or in services. No two ways about it. But when it comes to a specialist's touch, even just to verify whether the data is correct, we'll still be needed. Sure, the machine can make a mistake, the perspective will be different, or we won't have the full range of data, so AI won't take it over 100%. But we have to take it on the chin that our job as Ads specialists will evolve somewhat.

If someone was mainly a fan of clicking rather than talking to the client, or clicked without checking how it looks from a business perspective, that kind of job may already be dying out a bit. We're evolving precisely towards: let's ignore the clicks, see how it works as a whole, what the possibilities are and how we can improve something with the buttons available in Google Ads today. We have to learn to communicate, draw conclusions and check how our work affects the business, not just take the order, click it through, issue an invoice and survive another month. We have to become much more professional and broaden our skills. From my perspective, it's simply getting more interesting, because we're not a piece of the puzzle of the “I click here, but I don't really know what will happen” kind; we influence a bigger part of the process. That's exciting, but I also believe that for some people it may be too much, and they'll decide: “Okay, maybe let's change the direction of our career after all.” That may happen.

It's a bit like the threat to copywriters, right? AI writes texts, we write texts, it can do it better and more. In theory, yes, but at a professional level some pretty big differences come out. In our field, too, everyone can click. AI can click. But you still have to know what to click and what not to, which is also important, and then link it with the business, fuel or calm emotions on the business side, motivate someone or tell them not to do certain things in the business because they make no sense. So we still have this somewhat interpersonal element, soft skills, a bit of pseudo-psychological play: comfort someone where things are bad, lift someone up where things are better. So it's a bit of a game of emotions, data and moods, all at once. We've become a bit of a friend of the business, not just someone you call to click on something.

Which Google Ads campaigns should an online store start with?

Tomek: Absolutely, I agree 100%. I think it's very consistent with what, as in your copywriter example, is happening in other fields: AI most often replaces repetitive things. But the human factor is extremely important in purely advisory matters. You also added the psychological aspect, which I think is no less important. Karol, let's move on to something else we want to talk about in this episode, namely Google Ads campaigns specifically for e-commerce. Could you shed some light on which campaign types for e-commerce are likely to be the most effective, or which are a must-have for online stores in terms of setup? Where to start? We can even look at it from the perspective of someone just starting their business, for whom Google Ads is the first step they want to take to bring traffic to their store. Where would you start? What do you think is an absolute must, and what, on the contrary, would you not touch, especially at the beginning, so as not to cause the opposite effect?

Karol: Sure. Promoting online stores in Google Ads is a very repeatable scenario, so regardless of industry we run pretty much exactly the same campaigns. Let's start from the beginning. If we're starting out, the most important thing is, of course, to have a good product range and properly connected analytics. Without well-connected analytics we shouldn't go any further, because when after the first month we get asked what our return is, we won't have the data. That always has to be taken care of at the start. I've often seen someone, in the enthusiastic frenzy of “we're launching a new business”, decide: “Ah, we'll come back to analytics later.” And that “later” comes after the first month, when someone asks what worked and we have no idea. So let's just do it from the very beginning.

When it comes to campaigns dedicated to online stores, today we have two types in Google Ads: the Shopping campaign and the Performance Max campaign. Technically, at the very start, they don't differ much. We're still talking about the so-called product tiles that show up in search. Someone searches for our products, we show them, and that's basically what it's based on. The difference between these campaigns is the automation built into them. The Shopping campaign is entirely manual, meaning we set the click costs, and thanks to that we can calibrate bids and push better and worse products. Performance Max does that for us.

And here's the first thing we shouldn't do at the very start: we shouldn't begin with fully automated campaigns when we don't have data, because that's a bit like going to a casino. It may hit, it may not, but we won't be able to repeat it. The rule is simple. First we launch only and exclusively a Shopping campaign, calibrate all the click costs and check whether the campaign is profitable. If it is, we move to step two, which is automating it with a Performance Max campaign. If it isn't, we have to hit pause and check what we can improve in our business. What's interesting is that when that first, basic campaign doesn't work, people often tend to panic, switch on Performance Max and keep their fingers crossed that it'll somehow work out. That's more madness than method. I'd approach it more like a table: if the manual campaign works, let's automate it. If it doesn't, let's pause it and fix the fundamentals. We have to take a step back.

In a Performance Max campaign it's super important not to add so-called assets to it. Let me explain why. Technically, Performance Max is dedicated to online stores, meaning it shows those tiles in search. But if we add those elements, it suddenly starts showing on websites, on YouTube and in search in the form of other campaign types we have no control over. That will either waste our budget or pretend it's selling brilliantly, while in fact our company's revenue won't grow. In other words, we'll lose control, and it won't be a very pleasant story. So Performance Max, but without assets. It's worth checking and asking about, because a lot depends on this when it comes to whether our campaign succeeds or not. You can come back to it later, at some stage of growth, but at the very start it's key to skip it, because it's a risk. That one click can ruin us if we don't get it under control.

How do you split products in Performance Max, and what are zombie products?

Karol: The next step in campaigns for online stores is splitting Performance Max into smaller campaigns, taking into account the margin and profitability of the products we sell. Each campaign can have its own target return, and products have different margins. Once our business has grown in terms of the conversions we generate, we have to fit the account to the margins and product performance. We need to split products into, say, four typical groups. Our bestsellers always go into separate campaigns and get a bigger budget, to drive as much revenue as possible. The products that drag us down, the so-called cost drivers, are the ones that don't want to be profitable at a given point in their life. We should exclude them from our campaign plan or cut their budgets as much as possible, because we're burning our funds. Another category is the so-called steady sellers, because we always have a range that does okay. It's neither better nor worse, so we keep it separate.

And there's one more interesting product category we rarely look at. These are what we nicely named zombie products, meaning products the world has absolutely never seen. A small statistic that repeats in pretty much every e-commerce business: if a small store has a range of 1,000 products and we advertise all of them in one campaign that's profitable, the system will still focus on selling at most about 100 products and ignore the rest once it achieves results. In reality we have nearly 90% of products that nobody will ever see. And that's standard. Even though all these AI systems at Google work brilliantly, they'll never make the entire product range show up in parallel. To put it plainly: once the system smells blood, it focuses on what works, and better products may stay hidden from the audience.

The next stage is precisely catching these zombie products and giving them a fresh start, a new chance to be seen, because our next bestsellers may be among those hidden products. If they're not exposed, it's not so much that we're burning cash as that we're not making more of it. That's another task you can do month to month. And speaking of communication and reporting: a simple table. What percentage of products, and which ones, drive our revenue, which are only a cost and need to be minimised, and which are okay and can form the basis of our steady sales. And what share of zombie products we have, what's in there, and what the plan is to test it harder later and put it into ads. The icing on the cake is, of course, verifying the return from each of these categories separately, so we can steer them properly from month to month.

Do we expect higher revenue, or higher returns at the current budget? Here, too, we should decide month to month what we do with our campaigns, and we should get that as a recommendation from the contractor or ask for it. The question is fairly simple: can we increase budgets, or can we raise or lower ROAS? And most importantly: what will the consequences be, meaning what will happen in the business if we do it? A good standard would be to get, not in a report but in a recommendation, information along the lines of: we can increase the budget, ROAS will drop by only this much, for example, which will give us this business effect. Do we do it or not? Because you have to calculate whether we have the resources, whether we'll ship all those parcels and whether we even have the stock in the warehouse to handle it. That's exactly the exchange of information: what the system can guarantee us versus what state we're in offline, as a business, at the moment, and whether we can connect the two. That's the data we should be sharing.

When should you add Search, Display and Video campaigns?

Karol: Once we have that sorted, Performance Max campaigns in their various splits make up around 80–90% of the entire investment in e-commerce. Then add-ons come in. They may or may not, and that also depends on scale. So we add Search (text ad) campaigns. Those are the first four results we'll see below the Shopping ads. By nature, clicks in these campaigns are more expensive than in Shopping campaigns, so they work as a supplement, but the return from them will be lower. To put it plainly: if we can't invest any more in the best channel in Google Ads, we look for further elements. This is another element that may or may not work. We have to test it.

Later, once we have the customer acquisition part fully covered, we're left with campaigns for, as you might call it, the top of the funnel, meaning acquiring potentially new customers who haven't thought about our products at all. Here we have the world of banner, Display and Video campaigns at our disposal, meaning reaching people to inspire them to buy our products. It's worth noting that conceptually this works completely the other way round than on Facebook, so Google versus Facebook. In Google, when promoting e-commerce, we start from the so-called bottom of the funnel, meaning people who simply want to buy something from us, because they're the easiest people to present our product to, assuming our product is known to the public. Because if it isn't, this trick won't work. Then you have to inspire people. Have you bought anything interesting online recently? Something unusual?

What if the product first needs awareness, like ice-bath gear?

Tomek: Probably not. Although for me, a product that still needs awareness building is, for example, anything related to ice baths. Their popularity is obviously growing, and there really are more and more uses for them. Still, I sometimes get bombarded with ads, especially from foreign companies, and I can see they put a lot of effort into educating me about the benefits. I think that product would fit what you have in mind.

Karol: Yes. Let’s say that for an ice bath it would be wonderful to have designer foam briefs, so you don’t feel discomfort after getting into the cold water. And now the question is: does the world know these foam briefs exist or not? If it does, we follow the scenario we talked about. If it doesn’t, we have to start with inspiration, meaning Display and Video campaigns that say: “Do you take cold plunges? Maybe this product will change your life.” Then the concept would be reversed, but usually that’s a rare case.

Back to our order of campaigns. First, a Shopping campaign only, then automation with Performance Max without assets, then a proper split of Performance Max by margin and product performance. If we can’t pump any more money into that, then Search campaigns. When we hit the ceiling there, Display and Video campaigns inviting people to those cold baths. That’s how we can lay it out, of course.

I left remarketing out of this whole story, because remarketing is often built into Performance Max campaigns, so you rarely run an extra campaign, unless you have promotions, new arrivals, sales and so on – then you can add things like that. That’s a much smaller part. If we break it down, 80–90% really is product tiles, around 5–8% Search campaigns, and the rest is attempts with Display and Video to also try to cover the top of the funnel. But let’s not kid ourselves: Meta, for example, can often handle the top of the funnel better than Google. So here we work quite heavily in parallel.

The pattern actually repeats every single time, and most of the work sits in the middle part, meaning moving products around by their performance in Performance Max campaigns. At a larger scale, as we discussed at the beginning, you’ll need help here. If you run a small online store, one Performance Max campaign will be fine. You just enter the return you want, and it will run. But if you later need to split products by performance, the question arises whether you want to learn it yourself or outsource it. And who will that someone be? Who will be the contractor?

As a business grows, it’s not like you have to make this decision right at the start. It happens over time: you feel you’re short on either time or the skills to know what to click, and you have to decide whether to take it on yourself or hand it to someone. And if someone else, what the pros and cons are and what your expectations are, so you can pick the right person.

That’s really the whole magic know-how. It’s a method you can apply every single time. As you can see, it’s not secret knowledge at all; underneath it there’s a super uncomfortable part: someone has to monitor it week in, week out and react when something happens. It’s just tedious work, which of course can be automated to a large extent. Tedious, regular work that stands behind the success of online stores.

Do we need a brand campaign to defend against competitors?

Tomek: And brand traffic? Do we need a campaign here as well to defend ourselves against competitors?

Karol: We can use it as an add-on. I mentioned that this was a scenario for a brand that is only entering the market, so brand protection doesn’t matter, because at the start there is no brand. Later on, though, we can add brand campaigns. Let me point out right away that brand campaigns exist both in Search and in Performance Max, which is super important, because very often there’s this slip in the data: someone hasn’t excluded brand-related terms from the Performance Max campaign. And if the brand is strong, Performance Max can simply switch to showing on the company’s brand. In the table it looks great, but on the account it amounts to nothing.

Speaking of brand campaigns: we need a separate, brand-only Performance Max campaign, which can be done. Here’s a small trick. In all your sales campaigns, use the brand exclusions feature, and create one Performance Max campaign without it – that one will show only on brand terms. While you’re at it, it’s worth asking the contractor, or whoever runs the campaign, which part is brand and which part is sales, because that’s a completely different data analysis. And in Search we run a second, text campaign on brand terms for brand protection.

You just have to remember not to overdo the budget, because in most cases brand protection shouldn’t exceed 10% of the total budget. Otherwise you over-invest in protection, and protection doesn’t generate new sales, so it’s somewhat financially inefficient. There are situations where brand budgets are bigger, but that’s more of a nuance than a strategy. If you grow to the point where someone starts impersonating you – which is itself a sign of success, because someone wants to move into your field – then it’s worth protecting yourself.

It’s not worth counting this as results generated by Google Ads, because it’s only a defensive campaign. It doesn’t sell anything to anyone; it just protects you from losing a customer. It has to be split into separate campaigns. It can never be combined with your existing campaigns, because that distorts the picture and distorts optimization. And worse – remember when we talked about whether it’s worth paying on a revenue basis? Imagine you have a super strong brand and we launch a Performance Max campaign that only shows on that brand. In the table it looks great, but someone gets to cash in on the fact that we built a strong brand, which isn’t entirely fair. That’s how it could end in a really bad case.

What data beyond Google Analytics should the owner give the specialist?

Maciek: I’m wondering about one more thing. Beyond providing good analytics, beyond measuring the site for the main conversions and micro-conversions, can we as owners give the specialist any additional data from outside Google Analytics, so they can do their job better?

Karol: Yes, absolutely. That’s exactly the connection with data we can’t see in online tools. The first and most important thing: month to month, we should check whether what we see in analytics matches the business, because analytics can, to put it bluntly, break. Someone swaps the cookie popup, someone changes some code, and so on. So the first thing to check is what percentage of data coverage we have between Analytics and the booked figures, because those gaps mean we have to multiply certain things. That’s something to check month to month.

The second thing is returns or other things that meant a transaction didn’t happen. That ties in a bit with the first example. It may be that in Google Ads we see revenue of a million, and real revenue was a million, but 500 thousand came back. Our systems are fired up that sales are booming, and then it turns out that in fashion every second product was returned, and it no longer looks so great. We have to adjust our campaign expectations by this data too, meaning the so-called ROAS targets, because that’s the parameter we steer with.

Then there’s everything related to company costs. It would be good to keep track of those too, as far as you can share and exchange them, because as budget and sales scale up, there are more costs we don’t see in Google Ads, for example shipping, warehouses and so on. Those also affect what expectations we should have of the campaigns: whether we should raise or lower ROAS – no longer just relative to sales, but to the whole venture. That’s where the need for financial advice comes from, looking at how everything fits together overall, not just in this one channel.

And of course insights like: we’ve just brought this product in and it will have its launch, so there will be, to put it bluntly, more demand; or: this product is dying, we have the last units, let’s do something about it. Those are all things we don’t see in the data. Let’s call them product characteristics, offline or business ones, that you can’t come up with if you’re not in that industry.

Breaking it into categories again: number one is checking whether accounting and analytics are close to or far from each other, so we know the gaps and all the costs, based on which we have to work out what returns we need to get from Google Ads, or whether we can get them at all. That’s a matter of agreement. On top of that, insights that we, as internet people not connected to the industry, can’t see, while the person inside knows clearly what will or won’t happen with a given product in a given season.

Obviously, if there’s no snow, we won’t sell sleds. But if we have, say, running shoes, and the foam inside changes, and at some event that particular foam, and not another, is going to be heavily promoted, then as a keyword the running shoe is still white, but something completely different will be selling. So it’s better to position yourself for the upcoming trend. That’s also hard to come up with if, for example, you don’t run or you don’t get a big kick out of discovering these technologies. And here again comes the game with the contractor: do they know your product or not, or do they want to learn it? That’s the human side again. Or whether they even need to learn it at all. So we come back to: let’s talk about how we’re going to communicate and at what level.

How should you read GA4 data, and why does it never match the accounting?

Maciek: Definitely. What you just said, the information a specialist should expect from the business owner, is also proof that this conversation…

Tomek: …has to be, above all, two-way and really happen on a regular cycle, so that both sides make decisions more consciously.

Since we’re talking about decision-making: several times during today’s conversation, words and topics came up related to data analysis, data itself, and being well prepared and having the numbers worked out. That is inseparably tied to a tool that many people, unfortunately, currently hate, namely Google Analytics 4. You’ve already mentioned it. Of course we can’t expect the data flowing in there to always match reality perfectly, because we know there will be discrepancies.

First question: in your view, should the correct configuration rather be outsourced to someone who specializes in it, to get our measurement set up as well as possible? Second: is monthly measuring and comparing of data also meant to show whether a constant deviation between one set of data and the other is maintained, and possibly to alert us to anomalies? And third: what is your practice as a specialist when looking at data in Google Analytics? With what grain of salt, or what glasses do you put on, to read that data well and be able to talk about it with the store owner?

Karol: Yes, I put on glasses with the Google Ads logo, so everything looks better in my tables. Just kidding, of course. But starting from the very beginning: in Google Analytics, no data lines up with any other data. None at all. Let’s never expect it to match any of our accounting. We assume, starting with data volume, that if revenue in Google Analytics differs from what we have in the accounts by up to plus or minus 20%, we can work with that, so it already drifts quite a lot. Getting below 10% is a kind of rarity that sometimes can’t even be achieved. So right off the bat we have a margin of shift.

The minus in the data comes from the system not catching everything, while the plus can come, for example, from not having a system for analyzing returns. It can slip either way. If we have a shift on the plus side and we see that data in Analytics and in Google Ads, then imagine that… Let me use an example. A client has a 20% margin, so the expected ROAS should be roughly 500% and up. If we see revenue at a much higher level than it really is and set ROAS too low, we can kill our client, because we’re generating less revenue than needed, since the data didn’t match.

Sure, we have to implement analytics as well as we possibly can. The question is how far the store platforms we work on allow custom implementations and how closed they are, because that varies. But let’s agree: at least for purchases, we should measure as professionally as possible. No two ways about it. But plus or minus 20% – that’s the world we live in. It will be hard to reduce that in most small and mid-sized online stores.

We can add certain technologies, for example server-side tracking. That’s something that improves our results a lot. We’re talking about the so-called Google Tag Manager server-side, or solutions like Stape or TAGGRS, to improve quality. We’re also talking about tools like BigQuery, where we dump data from various channels and try to average it out there. But we still have to remember there will always be some deviation. These tools try to improve the quality of our data, but it will never be one to one. No chance. Unfortunately, we can forget about that.

You can simplify it to this: it doesn’t matter who does it, as long as it’s done as well as possible. Whether it’s a plugin, a paid solution, a specialist and so on – the most important thing is that it’s implemented and that it’s been checked how far it deviates from reality. So we implement the best available technology and remember that it’s no longer just Analytics, but also server-side solutions, everything related to BigQuery and so on. As far as we can, let’s put in the best solution available at the time.

Then there’s the question of using that data, because we calibrate our campaigns depending on those deviations. If, for example, the accounting data deviates from the data in Analytics, and we optimize for the data in Analytics, we have to build in a buffer. For example, setting a ROAS of 500 in Google Ads isn’t the same as a ROAS of 500 in the business. It plays out completely differently, and in the business it may be more or less, depending on the traffic channels we work with. Often someone tells us: “Okay, for every złoty I want to generate five in business terms,” and we, knowing the gaps between offline and online data, know that to do that we have to set the ad not at five, but at six or seven, to fill certain gaps in between.

That’s another thing that, unfortunately, not many people pay attention to: these are two different numbers. We have campaign expectations and we have business expectations, so we talk about business ROAS and campaign ROAS – and they will always be two things. If we set it too low in the campaign, the business will sink. If we set it higher, we can in turn reduce revenue, and that won’t be optimal either. That’s why periodic calibration and checking of deviations is necessary, because if something deviates in a given month and we don’t recalibrate, we’re behind for that week, those two weeks, that month when the system was working on bad data. It’s something that drifts and has to be controlled, which doesn’t feel very natural, because there’s an impression it should be one to one. But unfortunately these are two parallel worlds, so we have to check it.

Third topic: how do I verify it? Most often exactly as I said. At least once a month I always try to ask, or check if I have access to the store, what the real revenue is and whether there’s something there I don’t see – for example, that something happened somewhere else, someone placed orders under other types of orders or invoices, and so on. I ask about the whole business: how it went, whether any other costs popped up. So as to try to step a bit into the role of a financial advisor who uses Google Ads to help the business, and show where you can push harder and where it’s worth easing off.

When should a good specialist advise moving budget away from Google Ads?

Karol: Often it’s checking different channels at the same time, because not every contractor handling, say, Google, Facebook, TikTok, email marketing or anything else also checks all channels as a whole. Our role is sometimes to check how it all worked overall. There’s nothing wrong with saying: “Okay, dear client, Google Ads performed very poorly this month. Move the money from Google Ads to Facebook, because that will do a better job for you overall.” Sure, my commissions and so on may drop, but it’s good for the business. If I dug my heels in and kept suggesting “put more money into Google Ads” while the store’s overall financial result was lower, in the long run that wouldn’t work in my service’s favor.

Here you have to step out of your own world a bit and look at it like this: if we were the manager of this business, what would we invest in? The fact that we handle a certain slice shouldn’t affect that. And here we’d come back again to this: if we got a percentage of sales from a given channel, we could push suboptimal solutions, right? Sometimes it’s better to get a fixed fee for good advice so that everything grows than to make suggestions depending on what adds up for us at another level. It’s a pretty simple rule. The more money the client has, the longer we can sail on that ship with them. If the client sinks, we won’t issue another invoice. It makes no sense in the long run.

Our job is to keep it growing all the time. Whether with us or without us is another story. The key thing is that the client’s business keeps growing, even if we had to conclude that those briefs for swimming in cold water only sell on Facebook and not on Google. Let them sell on Facebook. All the best. I can help later as a contractor on analytics or drop the project, which is actually also great business-wise, because there’s a good chance that if the contractor steps back in a situation like that, saying it makes no sense, the client will be happy enough to recommend us to someone else. We come full circle a bit: we helped one business so much that we mixed ourselves out of it, but new projects appeared.

But you can be sure that if we advise shifting too much money into Google Ads and the business goes under, people won’t think very highly of us as specialists, and the other referrals simply won’t happen. You also have to look at it this way: sometimes losing a client can be something positive. The most important thing is that the loss comes from the business continuing to grow well. And if it’s without us – too bad. It’s a sad situation for a moment, but that’s not what it’s about.

We shouldn’t be that sentimental about projects, and that’s another interesting aspect worth discussing. Sometimes there’s a situation – I had this with my first projects – where I looked after them for a long time and in the first year grew them to the point where the client then took the work in-house. I was sad, because I lost a client and an invoice. But on the other hand, the business grew enough to carry it on. And that feeling that it’s my baby and so on – that sentiment is completely out of place, because someone actually paid for the time spent, so where does it even come from? We’re not shareholders, we don’t get a percentage of the business’s growth, but somewhere, as humans, it happens inside us. You have to cut those sentiments off; the numbers have to add up. That may sound a bit drastic, but it should be our foundation. It can be pleasant, but at the same time we can be poor, so maybe let’s make it pay off and be okay. They don’t have to be extremely positive or negative experiences.

What rules should a store owner take away?

Tomek: Karol, what you described I would call business maturity, and I’d add it as a fourth trait by which we judge a good specialist and agency when we’re looking for one. There’s so much truth and sense in what you say that I take it as a really huge argument for that kind of thinking and that approach. I’d like to look for exactly that kind of advisor myself – one with that much awareness.

It seems to me it all revolves around the communication you mentioned. We said that this communication happens on many different levels. In fact, it comes down to talking to each other, being honest, being open, and also being aware that you’re not just a contractor hired by the company to do some work. In a way, you really feel that you own this project and you’re responsible for it not only in your own little patch, but you can actually extrapolate it to the whole business. It’s been said for a long time that marketing is inseparably linked to business growth; it’s one of its main branches. Even if you’re an external advisor, in my opinion you automatically become a key figure in the organization. And again: if, as you said, you shut yourself away in your own world and are out of reach for the owners, or the owner is out of reach for you, then neither can you do the work effectively, nor can the owner be happy with you, because the flow of information is missing. I think that’s a nice bookend and the leitmotif of our conversation today: openness and communication.

Tell us at the end, Karol, if you can, because you really are a person with huge knowledge, and it’s clear how much you enjoy sharing it. This isn’t information we can read in any old trade magazine; you talk about what you actually experience and about your own observations. If you could, to sum up and put your seal on today’s conversation, leave our listeners with a few basic tips and rules, in your view, that they should follow when thinking about promoting their stores effectively through Google Ads or about working with agencies. Some bons mots you’d like to treat us to at the end – I think that would be a beautiful finale to our conversation.

Karol: I think it will be a summary in bullet points of what we talked about in this oh-so-short, nearly two-hour conversation. I’m a fan of asynchronous email communication, so doing bullet points and breaks at the end is something I love. A pure pleasure on my side. Maybe let’s start like this: to make the e-commerce world a better place.

First, let’s talk openly and transparently with the person running our campaigns about what the business data looks like as seen in the campaigns. As the opening of a conversation whose main theme should be whether we can invest more or get a better return on investment, and how we could do that. That’s a beautiful opening that will open the path to the next points.

Point number two is homework, meaning checking how the business data currently compares with our data in Analytics. Let’s try to see what gaps we have and check – this time with the contractor – what they may come from, so we can patch them as far as we can. And let’s try to make a first, super basic report in Excel where, based on data from Google Ads, Facebook, Analytics and other sources, we attribute how we could average out revenue across all channels, so we have a starting point. The task seems simple, but there will be lots of challenges along the way. It’s something we should be doing all the time, yet still few people even start it, so it would be a simple Excel sheet with performance. And yes, it will have to be outside Analytics, because some things have to be estimated.

The third point, also homework, is running or commissioning a product performance analysis for our e-commerce, meaning doing that split: what’s a bestseller, what drags us down, what’s a typical product and what’s a zombie product – so we can see how it’s distributed. Then we’ll see percentages like – I’ll take a guess – that in most cases up to 5% of products make 80% of the store’s revenue, while, for example, 30% of products drag us down. The logic is very simple. Seeing a chart like that, you just cut that 30% of unnecessary investment, and the world is already a much better place. Simply seeing that data makes things much better.

And the fourth element is the one nobody likes: practicing those three points regularly, at least monthly. It sounds like: you’ll feel better exercising more often, you’ll feel better eating better. Business-wise, you’ll also feel better having more control, doing mental exercise once a month to check how our products, our advertising channels and our returns on all investments are doing. So as not to start by doing everything – don’t change your life here and now by doing a million things, because nothing will happen. Just these three short points: talk to the contractor, or to someone in your team, about how we can sort out all the technical things, check the analytics, check product performance, and keep at it for at least six months.

That alone will trigger a few interesting moves. First, if there are gaps in the analytics, they’ll show up right away, because there will be lots of inconsistencies. You’ll also get lots of quick fixes out of it, because certain things will simply surface, for example investing in the wrong products, the wrong audiences, the wrong markets and so on. And if someone is a follower of the church of compound interest, such monthly corrections will deliver much better results by the end of the year. Our work, unfortunately, isn’t bursts of glitter and champagne, but percentages shaved off month after month, meaning small shifts that will add up very nicely by the end of the year.

Tomek: Great, beautiful. It really sounds like a ready prescription, a ready recipe. You also put it very nicely at the end, that tedious work that, in marketing terms, everyone often hides behind a nicer word like optimization. It sounds grand, but in practice it really comes down to doing something very regularly and very systematically.

Karol: We have a different shovel, but we should also dig regularly. The myth that our work is magical, secret and so on is a bit like: “Buy a diet pill, and tomorrow you’ll lose 50 kilos.” It sounds nice, but it may not work. We don’t have a diet pill, or a pill for higher profit, either. It’s exactly doing these unsexy things over and over, regularly, until suddenly, before we notice, we have a better summer body, or better test results, or we simply feel better day to day. If, for example, a doctor tells us “start hydrating,” we won’t pump 100 liters of water into ourselves on day one; we have to spread it out. It looks a bit like that for us too. We won’t increase sales tenfold from tomorrow, but we can multiply them over a longer period, only the jumps will rather be a few percentage points from period to period.

Where can you meet Karol?

Tomek: Great. Karol, thank you very much for today. It was an absolute pleasure to have you on our episode. A really nice and very substantive conversation, so once again, huge thanks. It’s great that you found the time to catch up with us. We know you sometimes give up your private time for this, so we appreciate it all the more. Since you regularly speak at events, I assume people can also meet you in person. So maybe, at the end, since I always ask our guests where it’s easiest to reach them. We already know you may not be keen on phone calls, so we certainly won’t leave your phone number here in the comments, God forbid. But is there a conference coming up soon where people will have a chance to hear you?

Karol: There are quite a few events. I don’t have them all in my head, but I have them all in my calendar. I post about everywhere I’ll be on my social media – though that’s too big a word. All the information about where you can meet me is on LinkedIn and on my website, so those are the two reference points. So that I don’t pull an unsure face because I can’t remember when and in which city I’m supposed to be, everything is simply collected in those places, because it’s offline talks, training sessions and so on. The calendar for Q4 is pretty full, so I’d rather refer to it than try to remember. I’m also not someone who’s impossible to reach. I may not pick up the phone, it’s true, but I’ll reply to a text or an email. If anyone would like to meet, even for the proverbial non-alcoholic beer, I’ll always let you know where I am. But the easiest is my website and LinkedIn. LinkedIn is quicker – all the information is there.

Tomek: Great. Karol, once again, thank you very much for the conversation. Have a nice day, and see you then.