Consent rate is one number hiding five problems

"Our consent rate is 44 percent." Fine, but that single figure tells you nothing about why. Are people rejecting on purpose, or bouncing before they decide? Is the drop on mobile or desktop? One region or all of them? A control variant or the new one? You can't fix what a single average hides. Measuring your banner properly means treating it as a funnel and reading each step.

To be clear on scope: this guide is about instrumenting your banner's own behavior, the interaction funnel. It's a different question from how much analytics data you lose to non-consent, which we cover in measuring the impact of consent on analytics.

The consent funnel

Every banner has the same underlying funnel. Instrument each stage:

Impressions. The banner was shown. This is your denominator. If it's lower than your sessions, you may have a rendering bug, or returning visitors whose choice is remembered (which is fine and expected).
Interactions. The visitor clicked something. Interaction rate (interactions divided by impressions) tells you how many people engage at all versus ignore the banner.
Outcome split. Of those who interacted: accept all, reject all, or customize. This is where the real story lives.
Ignored. Shown but never interacted with. A large ignore rate can mean the banner is easy to overlook, or that your setup lets people use the site without deciding (which has compliance implications).

Two sites with an identical 44 percent accept rate can be completely different underneath. One has a 90 percent interaction rate and people genuinely choosing; the other has half its visitors ignoring the banner entirely. Only the funnel shows you which one you are.

The metrics worth watching

Accept rate (accept divided by impressions). The headline number, but read it alongside the rest.
Interaction rate (any click divided by impressions). Low interaction often points to a visibility or copy problem, not a persuasion one.
Reject rate and customize rate. A high customize rate means people want granular control, worth honoring with a clear preferences layer.
Time to decision. How long between the banner appearing and the click. Very fast clicks can signal people swatting it away without reading, which matters if you care about informed consent.
Purpose-level opt-in. When visitors customize, which categories do they allow? Analytics might get 70 percent while marketing gets 30. That granularity tells you where the hesitation is.

Segment everything

An average across all traffic is where insight goes to die. The same metrics segmented reveal what to fix:

By device. Mobile and desktop often diverge, usually because the mobile layout hides the reject button or cramps the buttons. See mobile banner optimization.
By region. A GDPR-strict banner in the EU and an opt-out notice in the US will (and should) perform differently. Compare like with like. Benchmarks by region live in our consent-rate benchmarks.
By variant. If you're running an experiment, every metric above should be split by arm. That's the whole point of A/B testing your banner.
By page or source. A visitor landing on a checkout page behaves differently from one reading a blog post.

Account for signals you can't see as clicks

Some visitors decide before your banner ever asks. Browsers can send a Global Privacy Control signal (the GPC spec) that legally counts as an opt-out in several US states. Those users never click "reject," but they've rejected. If you don't count GPC, your reject rate looks lower than reality and your effective consent coverage looks higher than it is. A GPC-adjusted view of your numbers is the honest one.

Google's Consent Mode also models some conversions from users who denied consent, so your measured consent rate and your usable-data coverage are two different numbers. Know which one you're looking at.

Turn the numbers into changes

Metrics only matter if they drive a decision. A few common reads:

Low interaction, high accept among those who interact: the banner works for people who notice it, but too many don't. Fix visibility and copy, not the buttons.
High interaction, high reject: people see it and say no. Look at your purpose descriptions and whether you're asking for more than the visitor thinks is fair.
Mobile far below desktop: almost always a layout problem. Audit the mobile banner.
One region cratering: often a localization or translation issue rather than a real preference. Check the copy in that language.

Watch the trend, not the snapshot

A consent rate on a given day is noise. What matters is the line over time and what moved it. Chart your core metrics weekly and annotate deploys, because the most common cause of a sudden consent drop is a change someone shipped: a new tag that delays the banner, a redesign that shrank the reject button, a translation that broke. If accept rate falls the day after a release, you have your suspect. Watching the trend also catches slow erosion that a single reading never would, like a script that's gotten heavier month over month and now pushes the banner past the point where impatient visitors bail.

Where the data comes from

You can build this instrumentation yourself with custom events into your analytics, but the events have to fire before consent is granted, which is exactly when your analytics may be gated, so it takes care to do compliantly. A consent platform that records the funnel natively avoids that chicken-and-egg problem. CookieBeam's analytics dashboard tracks impressions, the accept/reject/customize split, purpose-level opt-in, and regional breakdowns out of the box, so you're reading a real funnel instead of inferring one from a single consent-rate number.