I understand the excitement. But I want to make an uncomfortable point: even if the Omnibus passes in its most marketer-friendly form, and even if every visitor consented to everything tomorrow, most companies would still make the wrong budget decisions. The problem was never just how much data we are allowed to collect. The problem is that we treat one tool as the truth, and we confuse attribution with impact.
The single source of truth problem
For fifteen years we have been spoiled with a free web analytics tool. Google Analytics became the standard not because it was the best way to understand marketing impact, but because it was free and everyone used it. Free became the norm, and the norm became the truth.
That truth has serious limitations, and GA4 made them more visible rather than less. It depends on consent and cookies, so it observes a shrinking, biased sample of your visitors. It is built around sessions and clicks, so it structurally favours the channels that harvest demand over the channels that create it. And its default reporting still pushes marketers toward last-click style thinking, which is a remarkably inaccurate way to value anything that works higher in the funnel.
The result is predictable. Companies that use GA4 as their single source of truth over-invest in bottom-funnel performance channels, because that is where the tool sees conversions, and under-invest in the mid and upper funnel activity that made those conversions possible. Branded search and retargeting look brilliant. Reach campaigns look worthless. Budgets follow the dashboard, and the dashboard is wrong.
My view is blunt: steering serious media budgets on GA4 alone is a risk most digitally mature companies can no longer afford. And connected to that, we need to let go of the idea that analytics should be free. Accurate, reliable data is one of the most valuable assets a marketing team can have. It is allowed to have a price tag. A paid analytics setup that gives you trustworthy insight pays for itself many times over in better allocation decisions.
What validation looks like in practice
If attribution cannot be trusted on its own, you need a way to validate it. Incrementality testing is one of the most reliable methods we have. Geo hold-outs and audience split tests answer the only question that really matters: what happens to my results when this campaign is not there?
In theory it sounds simple. Let me show you what it looks like in the real world, because the reality is messier and more instructive than the polished case studies suggest.
Earlier this year we ran a geo hold-out for a B2B lead generation client. We selected two Dutch provinces that had been highly comparable in lead volume for over a year, switched off all paid social in one of them for three months, and kept it running in the other.
The direction of the result was clear. In the three months before the test, the two provinces generated 46 and 41 leads. During the test, the province where paid social kept running dropped to 29 leads, while the hold-out province dropped to 17. Both declined, which tells you something else was going on as well, seasonality or market conditions. But the hold-out declined considerably harder. First indication: paid social was contributing real incremental leads, even though click-based reporting had always suggested the channel delivered relatively little.
Now the honest part. This test does not prove anything yet, and pretending otherwise would be exactly the kind of measurement theatre this article argues against. Budgets and campaign structures changed during the test period, so the comparison is not clean. The lead data came from a CRM export that included incomplete records, and part of the leads had no region attached at all. The client migrated CRM systems along the way. And we have not finished analysing the recovery phase, the months after we switched the region back on, which is where a hold-out test earns its final verdict.
There is one more lesson, and it might be the most important one. When the first results landed, nothing happened. No budget shifted, no KPI changed. The topic simply faded into the day-to-day, until a discussion months later about whether paid social was worth it brought the test back to the table. That is not a failure of the test. It is a failure mode of organisations, and I suspect every reader recognises it.
Four lessons from an imperfect test
First, keep everything else constant. An incrementality test is only as clean as the environment around it. If budgets, campaigns or landing pages change mid-test, you are measuring noise.
Second, fix your data hygiene before you start. If your lead source cannot reliably tell you which region a lead came from, a geo test will always end in caveats.
Third, measure the recovery. Switching a region back on and watching whether volume returns is half the evidence, and it is the half most teams skip.
Fourth, agree upfront what decision the test will trigger. A test without a pre-committed decision rule produces an interesting slide, not a change in behaviour.
And a note for B2B marketers specifically: at 40 leads per quarter, statistical certainty is a luxury you will not get from a single test. That is not a reason to skip testing. It is a reason to test continuously, read results as evidence rather than verdicts, and combine them with other methods.
Where measurement goes next
Those other methods are maturing quickly, and this is where I am genuinely optimistic.
On the tactical level, the future of user-level measurement is not the cookie. There are far more signals hiding behind a user than a cookie ever captured: anonymised IP, browser and device characteristics, local storage, cache, behaviour patterns, and logged-in environments such as customer portals. Modern measurement platforms lay an AI model over these signals to match users across sessions and devices probabilistically. That is what makes it possible to estimate the value of campaigns that were never designed to be clicked, the impression-driven, reach-oriented work that click-based tools have always undervalued. Only then can you judge campaign value at a tactical level and actually act on it.
On the strategic level, marketing mix modelling is having a second life, and deservedly so. It is an old methodology, but the tooling has become dramatically more accessible, with Google actively pushing its open-source Meridian framework. MMM is a powerful way to understand the long-term value of above-the-line and brand investment, and to buy media smarter over time.
The end state I advise clients to work toward is triangulation. Attribution for daily tactical steering, incrementality testing as the continuous validation layer, and MMM for strategic budget allocation. No single method is the truth. The truth lives in the overlap. And underneath all three sits the least glamorous asset of all: first-party data, collected in a robust CRM through email, on-site behaviour and purchase history, because models are only as good as the data you feed them.
So what about the Omnibus?
Back to where we started. I think simplifying consent is a sensible step. Cookie banners in their current form are broken: endless walls of options that almost nobody uses as intended. Nobody should mourn them.
But two things need to be true for this to end well. The proposal has to actually land, and right now that is far from certain, with core consent provisions already dropped in the Council’s own text. And the industry has to adapt fast, because today’s entire web analytics and tracking stack is built around the current consent model. Advertising platforms, CMPs and analytics vendors will need to move in step with whatever new definitions of consent and cookie categories emerge. The fewer grey areas regulators leave, the better companies will comply, and the brands that adapt fastest will collect more and better data than their competitors while everyone else is still reading legal memos.
Just do not confuse that with a measurement strategy. Regulation determines how much data you may collect. It says nothing about whether you interpret it correctly.
My advice to any digitally mature company is therefore independent of what happens in Brussels. Stop treating one free tool as the truth. Build a validation layer with continuous incrementality testing, and accept that your first tests will be imperfect. Add MMM for the strategic picture. And treat accurate data as what it is: an asset worth paying for. Companies that do this will make better decisions in every regulatory scenario. Companies that wait for the Omnibus to solve their measurement problem will discover it never could.
Ascend Online is sponsor of DDMA Digital Analytics Summit 2026. More information and tickets for Digital Analytics Summit taking place on 1 October at Hotel Casa in Amsterdam.
Ryan Peerkhan
Ook interessant
The consent debate is a distraction from the real measurement problem
What does it really cost to own an analytics platform? Software, migration, and maintenance costs explained
