For years, we lulled ourselves into the security of “we can measure everything.”
Who saw the ad. Who clicked. Who returned after three days. Who bought. Which campaign generated the lead. Which platform brought the conversion.
Then at a certain point, the gravy train stopped.
Browsers limited cross-site tracking, users started managing consents, ad systems capped cookie lifespans, and mobile devices began sending us fewer and fewer pieces of information. In short: fewer and fewer data points available to read and analyze.
The documentation published on MDN regarding third-party cookies explains that Safari and Firefox introduced default protections against cross-site tracking. Google, in its October 2025 update on Privacy Sandbox, confirmed its intention to maintain a user-choice-based approach regarding third-party cookies in Chrome and announced the retirement of several planned APIs.
Meanwhile, users are more attentive. According to Eurostat, in 2025, 76.9% of internet users in the European Union took measures to manage access to their personal data online. 58.8% chose not to allow the use of their data for advertising purposes.
Cookie banners tell us something too. The Didomi 2026 benchmark on consent rates in Europe, based on 2025 data, indicates opt-in rates between 55.7% in Western Europe and 67.6% in Eastern Europe. A significant portion of traffic does not produce clean and convenient signals like dashboards would prefer.
At that point, many companies started saying: “Oh dear! With privacy, you can’t do marketing like you used to.”
True. But only in part.
The useful question isn’t: how do we recover all the data we had before?
The useful question is: when we had all that data, were we actually using it to make better decisions?
From experience, in 90% of cases, the answer is NO! You had the data before and you weren’t using it. So what do you have to complain about?
Besides, we haven’t gone from a world where we knew everything to a world where we know nothing. Tracking still exists; it’s just that instead of measuring “everything,” you have to use your head and figure out what you need and how to get it.
So yes, the problem exists. But it’s only the premise.
IAB 2026: measurement is under pressure
Let’s start with a few numbers so that, since we’re talking about data, we don’t get things wrong: the IAB State of Data 2026 report starts with a clear picture: privacy regulation, signal loss, in-platform optimization, and data environment fragmentation make it harder to connect media exposure with real results.
Precisely for this reason, IAB launched Project Eidos, an initiative to modernize advertising measurement. In Europe too, the topic is tangible: the IAB Europe report on addressability and measurement tells us that access to cross-platform data is the main challenge for the industry, cited by 68% of respondents, followed by privacy regulations at 58% and signal loss related to cookies at 48%.
These numbers say one simple thing: marketing hasn’t become impossible. It’s just become less comfortable.
To simplify:
The data is there.
- You just have to choose which ones you want to collect and stop spamming random tracking pixels.
- It’s fragmented, so you need to connect and assemble it.
- Sometimes it’s not quantitative data, but qualitative signals that you need to be able to intercept, relate, and interpret!
Let’s get real! Masks off, and let’s talk about what actually happens inside companies: throwing around tons of tracking pixels, analytics events, 200 audience segments, and 400 dashboards serves absolutely zero purpose. In daily operations, that stuff only serves to create reports and presentations full of numbers and charts that no one reads.
When you collect lots of data without method, you don’t have more control. You have more noise. Instead, you need to use your head and understand which data points are needed to create assets for evaluations and decisions, rather than noise!
The alibi
Privacy becomes annoying because it removes an excuse we used to hide behind.
When tracking was more abundant, it was easy to tell ourselves that gathering everything would be enough, and then—sooner or later—someone would figure out what to do with it. Spoiler: that doesn’t happen.
Privacy-first marketing forces you to ask a grown-up question: which decision should this data improve?
If you can’t answer, then maybe you don’t need that data. Or maybe you’re asking the wrong person, at the wrong time, without giving anything in return.
And here comes the GDPR principle of data minimization. The European Commission reminds us that personal data must be adequate, relevant, and limited to what is necessary in relation to the purposes for which they are processed.
Put without legal jargon: don’t collect data because “it might come in handy.” Collect it because it actually serves a clear purpose.
First decide what you need, then collect the data
The rule is simple: first the decision, then the data. Not the other way around.
Want to know which leads to work on immediately? Define what makes a lead priority: urgency, company size, sector, declared need, person’s role, requested product, recent behavior. You don’t need to know everything. You need to know enough to prevent a warm contact from cooling down.
Want to know which campaigns actually work? Cost per lead isn’t enough. Connect that lead to the sales pipeline, customer lifetime value, CRM information, margin, and quality of the generated customer.
Want to improve retention? Chasing the user across half the web makes no sense if you can’t distinguish a loyal customer from someone who hasn’t bought in eighteen months. You need data on purchase frequency, open tickets, complaints, and reasons for churn.
Before adding a field to a form, configuring an event, or creating a segment, ask five questions: what decision does this data improve? Who uses it? When? What action does it trigger? What improves for the person sharing it?
Asking for product preferences in an e-commerce store makes sense if you then send relevant content, offers, or suggestions. It makes no sense if, after sign-up, you still send the same generic promo to the entire database.
What happens if you don’t do this
If you don’t do this work, costs arrive—even when you don’t see them right away.
You burn budget optimizing campaigns on weak conversions. You bring in leads that only look good because they cost little. Sales works them, wastes time, and the usual war between marketing and sales starts: “we bring in the leads,” “yeah, but they’re garbage.”
You lose internal trust because every dashboard tells a different story. Meetings become arguments over which data point is the right one, instead of what action to take.
You lose customer trust because you ask for data and then misuse it. The user fills out preferences and receives spammy generic communications. They buy a product and keep seeing ads for that exact product for weeks. That’s not personalization. It’s annoyance attached to their name.
And AI?
The IAB State of Data 2026 report also addresses AI’s role in measurement: attribution, incrementality testing, and marketing mix modeling. It makes sense. If signals are more fragmented, more sophisticated tools are needed to interpret incomplete data.
The problem is thinking AI can fix messy data, poorly updated CRMs, conflicting definitions between marketing and sales, or segments created once and never reviewed.
If a model sees many conversions, but those conversions are off-target leads or contacts sales discards after thirty seconds, AI isn’t finding value. It’s optimizing garbage.
AI doesn’t transform useless data into strategy. It just packages it better.
So obviously AI can help us immensely! If we don’t use it to process huge amounts of data, what else should we use it for? But if you put poor-quality ingredients in a pot, the final dish is guaranteed to taste terrible! AI needs to be fed the right data, properly packaged and organized; agents must be well constructed. Otherwise, you’re just dumping garbage into the pot and hoping a gourmet meal comes out.
That doesn’t happen—you just end up with cooked trash on your plate!
Conclusion
So how do I conclude this article? What do you want me to say? Instead of keeping the same pre-privacy mindset of “let’s measure everything and figure out what to do with it later” while cursing privacy rules, let’s use privacy as a engine driving us to work better. Figure out first “what information I need” and then work backward to determine what to measure and how. Then put effort into data quality instead of throwing everything into Claude to see what happens.
And step by step, you’ll see that privacy will no longer be a problem, numbers will start making sense, and you’ll work better with less clutter on your desk.
