If your Google Ads or GA4 remarketing audience — or your Meta Custom Audience — has been quietly shrinking, the instinct is to blame list decay or a slow traffic month. Sometimes that's right. But if the drop lines up with a Consent Mode rollout, a new cookie banner, a CMP swap, or an iOS update, the actual cause is often consent gating working exactly as designed — visitors who deny consent are, by design, no longer eligible to be added to a remarketing list. The question worth answering isn't "is this normal," it's "is this the expected effect of consent gating, or a misconfiguration quietly undercounting people who did consent."
1. Line up the drop against your Consent Mode timeline first
Before touching DevTools, check the date. Open Google Ads' Audience Manager or GA4's Audiences report, look at the size trend over the last 3-6 months, and compare it against when Consent Mode, a new banner, or a CMP migration actually went live. A drop that starts the same week as a consent rollout is a strong signal it's related. A drop that's been gradual for a year, unrelated to any consent change, is more likely ordinary list decay or a traffic problem — don't skip this step, it's the difference between "working as intended" and "something to actually fix."
2. Check whether Consent Mode is running Basic or Advanced
This matters because the two modes affect your remarketing numbers very differently. Basic mode blocks Google tags from firing at all until consent is granted — no tag, no audience membership, full stop. Advanced mode lets tags fire immediately but sends cookieless pings and uses conversion modeling to estimate results for consent-denied traffic. Modeled conversions can make your reporting numbers look healthier even while your actual addressable remarketing list shrinks, because a modeled conversion isn't the same thing as a real, taggable visitor you can re-target.
How to check: same DevTools process as checking whether cookies fire before
consent — open the Network tab, look for
requests to google-analytics.com or googleads.g.doubleclick.net, and check for a gcs
parameter in the URL. Present means Advanced mode's ping behavior; absent (with full
payloads going out) means something's misconfigured, not intentionally modeled.
3. Rule out a Consent Mode implementation bug, not just the policy itself
A genuinely correct Consent Mode setup shrinking your audience is consent gating doing its job. A broken Consent Mode setup can shrink it further than it should, by failing to register consent from visitors who actually clicked "Accept." This is a different problem with a different fix — and it's common enough that it's worth ruling out explicitly rather than assuming the drop is purely policy-driven.
How to check: in DevTools console, run dataLayer right after clicking Accept on your
own banner and look for a consent entry with "update" and all relevant categories set
to "granted". If that update event never fires — or if the default was stuck on
"denied" before it — visitors who genuinely consented aren't being counted, and your
audience is smaller than it should be. The full breakdown of what to check, covering both
the default-state and update-event failure modes, is in why Google Consent Mode v2 isn't
working.
4. If it's a Meta Custom Audience, separate consent gating from iOS ATT
Meta's Custom and Lookalike Audiences shrink for a related but distinct reason: Apple's App Tracking Transparency (ATT) opt-in prompt, on top of whatever your own cookie banner is gating. A visitor who accepts your site's consent banner but declines the iOS ATT prompt (or has "Allow Apps to Request to Track" disabled system-wide) still won't be trackable by Meta's pixel — that's an Apple-level restriction your consent banner has no control over, and it's easy to misattribute to a consent-gating bug that doesn't actually exist.
How to check: in Meta Events Manager, check the Aggregated Event Measurement diagnostics for the pixel and compare match rates on iOS traffic against Android/desktop. A large gap specific to iOS points to ATT opt-out rates, not your consent banner. If the drop is even across all platforms and device types instead, that's a signal to go back and check your own Consent Mode / CMP setup as described above, since ATT can't explain a platform-uniform drop.
5. Check whether this is a regulator-driven cutoff, not just your own setup
Google has, in specific markets, cut off certain conversion tracking signals entirely for advertisers it judges non-compliant with consent requirements — this isn't hypothetical, it's happened in the EU tied directly to Consent Mode compliance: starting July 2025, Google began disabling conversion tracking, remarketing, and ad personalization for EU-traffic advertisers found non-compliant with its Consent Mode requirements.
How to check: in Google Ads, check for any account-level policy notices or restricted status related to the EU User Consent Policy. If your traffic is EU-heavy, the drop is sharper than a typical implementation issue would explain, and steps 1-3 above didn't turn up a clear cause, this platform-level enforcement is worth ruling in or out before assuming the problem is purely on your side.
If Consent Mode is missing ad_user_data or ad_personalization specifically
A related but distinct check: even a Consent Mode setup that fires correctly can be
missing the ad_user_data and ad_personalization consent types specifically — the two
signals added in the move from v1 to v2, and the ones Google actually needs to add
visitors to a remarketing audience or use their data for personalized ads. A default or
update call that only mentions ad_storage and analytics_storage (the original two) can
look entirely correct while still leaving these two newer signals unset. See Consent Mode
v2 missing ad_user_data and ad_personalization
for how to check for this specifically and fix it.
What a smaller-but-real audience actually means
A remarketing audience that shrinks because consent gating is working correctly isn't a bug to fix — it's a smaller, cleaner list of people who can actually be legally re-targeted, which tends to perform better per-visitor even if it converts fewer people in raw volume. The goal of this check isn't to claw back every visitor who said no. It's to make sure you're not also losing the visitors who said yes, to a misconfiguration nobody's checked.
If you just want the implementation checked
Steps 2 and 3 above — whether Consent Mode is set up in Basic or Advanced mode, and whether the consent default and update events are actually firing correctly — are exactly what a free pre-consent scan checks automatically, so you're not doing the dataLayer digging by hand every time you suspect a drop.
Petrus, Founder — ConsentCheck. I've chased this exact "why did the audience shrink" question for clients more than once, and it's almost always one of the five things above.