Who really profits from your ads?
Meta shows its results. Google shows its own. Google Analytics shows yet another set. The problem starts when you try to decide, based on these numbers, where it's worth spending the next few thousand złoty.
In our conversation with Mariusz Michalczuk of Conversion, we look at why data from ad platforms can lead marketers astray. Each system has its own way of attributing sales, so the same transaction can be credited to both Facebook and Google at the same time. On screen, everything looks pretty good. Only when you compare this data with the actual business result does it turn out that the numbers don't always tell the same story.
Watch the conversation (in Polish).
Full episode of the SUURowo o marketingu podcast.
Every platform has its own truth
Attribution sounds like a term from a presentation for analysts, but its consequences are very concrete. If Meta credits itself with sales according to one model, Google Ads according to another, and GA4 interprets the user journey differently again, it's hard to expect all the reports to show the same picture. That's exactly why adding up the results from individual platforms can lead to absurd conclusions.
Mariusz points out that the problem isn't that one of these platforms is necessarily "lying". Each of them looks at the data through its own measurement system. The problem arises when we treat these reports as one objective truth about the business. If we shift budgets between channels on that basis, we may start investing more and more in activities that look great on the dashboard but much worse in the company's actual results.
An analytics mistake can cost more than a campaign
The most dangerous mistakes aren't always visible straight away. The conversation includes the example of an e-commerce company that was losing around 17,000 dollars a week because of incorrectly configured elements. With large sales volumes, everything could look normal at first glance. Only a closer look at the data showed where the money was really disappearing.
This shows well what analytics should be in practice. It's not just about collecting as much data as possible and having more and more charts. Mariusz breaks the process down into several elements: you need to collect the data correctly, ensure its quality and stability, be able to analyse it and, above all, turn it into business decisions. A report alone doesn't make money. Only the decision you make based on it does.
Data is getting harder
Just a few years ago, a marketer could feel that with the tools set up properly, they would see almost the entire user journey. Today that certainty is fading. Changes related to privacy and tracking restrictions in browsers mean that some information simply no longer reaches the systems the way it used to.
That's why server-side tracking is becoming more and more important. It gives a company more control over what data is sent to Meta or Google, and at the same time lets you combine marketing data with information the ad platforms don't have, such as margin or actual sales results.
And here is another important change: it's not always worth fighting obsessively for 100 percent of the data. Mariusz points out that analytics can show around 80-85 percent of traffic and still provide valuable information. More important than completeness itself is whether the data is of good quality and whether we understand its context.
AI won't fix bad data
Artificial intelligence is also entering the world of analytics more and more boldly. Tools that can analyse a report and immediately point out potential problems sound like every marketer's dream. But here, too, it's easy to get a false sense of security.
Mariusz talks about his experience with AI in Google Analytics and situations where the tool drew wrong conclusions or simply made something up. For someone who knows the data well and can verify the answer, it can be a useful assistant. For someone who takes its interpretation as certain without checking, it can become yet another source of wrong decisions.
More and more traffic may not end in a click
It's not only the way data is collected that is changing, but the internet itself. Users increasingly use LLMs and generative AI tools instead of going through the traditional path: search engine → website → article → form.
Mariusz already sees traffic from LLMs at a few percent among his clients, roughly 3-5 percent. The consequences of AI Overviews are even more interesting. If a user gets the answer directly in the search results, they may not visit the website, sign up for the newsletter or be reached by remarketing. From the company's point of view, this isn't just a change in the source of traffic. It's a change in the whole way you can build a relationship with a potential client.
Not every conversion problem starts with an ad
The conversation also raises an important point that's easy to forget when we look at all marketing through the lens of campaigns. Sometimes the ad works, the user lands on the website and is interested in the product, but still doesn't buy. The reason can be trivial: no information about delivery costs, an overly complicated checkout or another element that only appears at the end of the purchase path.
One simple way to find such problems is the so-called corridor test. Just ask someone who doesn't work in marketing or technology to try shopping in your store, and watch where the difficulties arise. Quantitative data tells us where users drop off. Observation and conversation help us understand why.
This approach shows well why quantitative analytics alone isn't always enough. We may know that 40 percent of users drop off at a particular stage, but only qualitative data can tell us what is actually stopping them.
One source of truth instead of another dashboard
If every system shows a slightly different version of reality, the solution isn't to add another tool that shows yet another report. What you need is one place where data can be combined and related to what the company actually cares about.
In practice this means building your own source of truth, based on business data rather than only on what a particular ad platform reports. Then you can check whether a change in budget really translated into results, instead of looking only at the growth in conversions attributed by Facebook or Google.
A good example is the PDCA approach: plan a change, implement it, check the result and decide what to do next. If you move part of your budget from one channel to another, the most important question isn't "which dashboard shows more conversions?", but "what happened to the result of the whole business?".
How to approach analytics so you spend your budget better?
Listen to the full conversation with Mariusz Michalczuk (in Polish) if you spend money on advertising, run e-commerce or are responsible for marketing and want to know which activities actually deliver results for your company.
Want marketing that works for your business?
We design and deliver B2B marketing - from strategy, content and social media to personal branding and marketing outsourcing.