Most brands treat marketing consent (consent mode) as a reporting issue. „Part of the revenue has no attributed source, we have gaps in attribution.” That’s true, but it’s the least expensive part of the problem.
The real cost shows up earlier, on the campaign side.
Fewer consents mean fewer signals for the algorithms
When a user declines tracking, it isn’t just a row that disappears from a report. What disappears is a signal that would otherwise reach Meta, Google or TikTok: information about a purchase, a cart, a conversion.
Ad algorithms learn from those signals. The fewer they receive, the worse they optimise: weaker audience targeting, harder budget scaling, slower reduction in cost per conversion. The chain is simple:
fewer consents → fewer signals → weaker algorithm learning → worse optimisation and lower ROAS.
As a result, consent mode errors are not a reporting problem but a problem of ad spend efficiency. The gaps in reports are only the visible symptom.
Legal compliance is an obligation, the way it’s implemented is a choice
Respecting consent mode is mandatory. Without consent, data cannot be sent to ad platforms, and there’s no room for debate there. The room for decisions is elsewhere: in how well the entire consent-collection process is designed.
The same legal requirements can be met in a way that loses a few percent of signals, or in a way that loses more than ten. The difference doesn’t lie in the law, but in the banner UX, in how the choices are constructed, and in whether anyone is checking that the mechanism actually works.
And this is where most companies have a blind spot. They don’t see the scale of the loss, so they don’t know there’s anything to fix.
What we changed in Sublime
Until now, all unattributed revenue landed in a single bucket – „untracked”. We split it into two categories:
- untracked – no signal of any kind, no event was matched,
- consent denied – a purchase event exists, but the user did not consent to tracking.
As a result, the attribution dashboard shows two distinct things directly: how much revenue is genuinely untracked, and how much is lost specifically due to missing consent.
This distinction matters because it leads to different actions. High „untracked” is usually a tracking-configuration problem. High „consent denied” is a consent-collection problem, and at the same time a measure of how much data isn’t reaching the campaigns.
How much this really is
Across our clients’ data, on average around 7% of revenue falls into the consent denied category.
More important than the average is the spread. The differences between clients are large, which means the level of loss isn’t a market constant independent of implementation. It depends on configuration, and is therefore fixable.
Consent denied as an indicator of signal health
A single measurement says little. The value appears when the category is observed over time.
Banner changes, a CMP update, a rebuilt purchase path – each of these can quietly degrade consent collection. In campaign results this shows up with a delay, once the algorithms have already reacted to the poorer data stream. In the consent denied category it shows up immediately.
This is where Sublime’s role in this area lies. Optimising the consent process itself depends on the banner, the CMP and the UX on the store’s side, and Sublime does not take that over. It takes over something else, something no standard tool does today: continuous monitoring of this category and an early warning the moment consent collection starts to break down – before the losses grow and before they show up in campaign results. That turns consent denied from a number in a report into an early warning system.
What can be done on the store’s side when this indicator rises:
- clearer banner UX: contrast, no obstruction of an informed choice,
- a simpler decision for the user instead of overwhelming them with options,
- re-asking for consent (consent refresh) for returning users,
- treating consent as a growth lever, not only a compliance topic.
A different question to close
The standard question is: „how much revenue do we have unattributed?”. That’s a question about a report.
The better question is: „how much signal isn’t reaching the ad algorithms?”. That’s a question about campaign results, and it’s the one consent denied answers directly.
