Most businesses in Europe installed a cookie consent banner, lost a slice of their measurement overnight, and then carried on reading their reports exactly as before. At The Growth Bully, a Malta performance marketing agency, that is the most common measurement fault we find in an account we inherit, and it has usually been quietly wrong for years.
The banner did not break your marketing. It broke your evidence, and nobody updated the paperwork.
What did the consent banner actually change?
It changed who can be measured, not who can be marketed to. Visitors who decline analytics and advertising storage still arrive, still read, still enquire and still buy. They simply stop appearing in the data your decisions rest on, so every rate you calculate now comes from a smaller sample than it used to.
That matters most where numbers are compared across time. A conversion rate from before the banner and one from after it are not the same measurement, and the step between them gets read as a performance drop rather than as a change in what was being counted.
What is consent mode and what does it do?
Consent mode is a signal layer. It tells the tags on your site whether a visitor agreed to analytics and advertising storage, and the tags then behave differently depending on the answer. In its basic form nothing fires until consent is given. In its advanced form tags still send a cookieless signal when consent is absent.
The practical difference is large. Basic consent mode leaves you blind to anybody who declines. Advanced consent mode passes an anonymous, non identifying signal that something happened, which is what later allows a gap to be estimated rather than left empty. Both are legitimate configurations. Which one is appropriate for your business is a question for whoever advises you on data protection, not for your media buyer.
Why does your conversion count look lower than it used to?
Because a share of your conversions now happens outside what the tags are allowed to observe. Nothing was lost in the business. The record of it was. If a meaningful share of visitors decline, you should expect a permanent shortfall between platform reported conversions and the orders in your own system.
The test that separates a measurement gap from a genuine decline is the one in why analytics and ad platforms disagree. A stable shortfall is two systems working as designed. A shortfall that moves around month to month is a broken implementation, and that one is worth fixing this week.
What is modelled data and should you trust it?
Modelled data is an estimate. Where consent is missing, platforms use the observed behaviour of consenting visitors to infer what the rest probably did, then report the result next to measured conversions. It is useful for direction and it is not a count. Presenting it as a count is how reporting becomes fiction.
Treat a modelled figure the way you would treat a survey result. It is good enough to decide which campaign is working better than another. It is not good enough to reconcile against an invoice. If your reporting mixes measured and modelled conversions in one column without saying so, that column cannot settle an argument about revenue.
What does server side tracking actually change?
It changes where data is processed, not whether you are allowed to process it. Tags fire into a server you control, that server decides what is forwarded onward, and the browser does less of the work. The gain is durability and control. The permission still has to come from the visitor.
Durability is worth having on its own merits. Browser side tags are the most fragile part of any measurement setup and they break whenever something changes in a page, an extension or a browser release. Moving the decision about what gets sent into one controlled place is the same instinct behind conversion tracking that survives, and it is a resilience measure rather than a clever route around the banner.
Is the conversions API a way around consent?
No, and it should never be sold as one. Sending conversions from your server rather than from the browser is a change of transport. The obligation to hold a lawful basis for sending that data, and to respect a visitor who declined, is unchanged by the route it travels. Anyone describing it as a loophole is describing a liability.
What it does do well is reduce loss. A conversion confirmed in your own system, forwarded once, with consent respected, is more reliable than one that depended on a browser pixel firing at exactly the right moment. That is the whole case for it, and the case is strong enough without pretending it does anything else.
What should you fix first?
Start with the honesty of the reporting rather than the volume of the data. Most accounts do not need more tracking. They need the numbers they already have labelled correctly, so that decisions stop being taken against a figure nobody in the room can explain.
- Find out what your real consent rate is. Everything downstream is a function of it and almost nobody knows their own number.
- Label measured and modelled separately in every report, in their own columns, so an estimate is never quoted as a count.
- Re baseline from the day the banner changed. Comparisons that cross that line are not comparisons, and they should carry a dated note explaining the step.
- Make your own system the decision maker. Enquiries and closed revenue in your customer relationship system are the one record consent cannot thin out.
- Check the banner does what you were told it does. Declining should visibly stop tags from firing, and verifying that takes about five minutes.
What should you judge once you accept the gap?
Trends in cost per qualified enquiry, measured inside your own account, on a setup you have not changed. That question survives a consent change, a browser release and a platform update, because your own records answer it rather than a pixel that may or may not have been permitted to fire.
It also makes the method matter less than most people assume. Once you are judging your own trend line, arguments about attribution windows become arguments about settings rather than about reality, and the thing being optimised is revenue you can actually see. That is the standard we hold every digital marketing account to, and it is what the Pipeline Scorecard is built to expose.
If your reporting has not been re baselined since the banner went up, next month is being planned against a number that stopped being true some time ago. Tell us what you are measuring and we will tell you what it is missing.