Consent Mode v2 adds two signals on top of the original two: ad_user_data and ad_personalization now sit alongside ad_storage and analytics_storage. The version arrived with Google's November 2023 update, and on sites that serve EEA visitors it is now part of keeping Google Ads measurement and personalization working. The practical question is not whether you installed it, but whether it is actually firing.
This guide walks through verification in order. You will learn what each signal controls, how the default state differs from the update that fires when someone clicks your banner, where to see the state in Tag Assistant and GTM Preview, and how to read the gcs parameter in a network request. It closes with a monitoring routine and a grace period so you do not panic over normal latency.
The path in 6 steps
The steps of this page, in order.
-
01 Know the signals Understand what each of the four consent signals controls before you test anything.
-
02 Set your default Know what state the page should be in before the visitor touches the banner.
-
03 Confirm in tools Use Tag Assistant and GTM Preview to see the state the page reports.
-
04 Read the request Inspect the gcs value in network requests to prove signals leave the browser.
-
05 Debug failures Match common symptoms to causes before blaming Google.
-
06 Monitor after launch Watch the signals weekly and give yourself a grace period before calling it data loss.
01 / 10
What are the four consent signals in Consent Mode v2?
Four switches control everything: ad_storage and analytics_storage from the original version, plus ad_user_data and ad_personalization added in November 2023.
Each of the four signals, ad_storage, analytics_storage, ad_user_data, ad_personalization, is a binary switch Google tags read before acting. The first two govern cookie writes: advertising identifiers versus analytics cookies. The last two govern data flow, not storage.
All four operate on a simple granted/denied value, so verification always checks the state, never a percentage. Basic and advanced modes change how those states are transmitted, not what the signals mean.
Ad_storage
Controls whether Google tags may write advertising cookies and identifiers on the device. When it flips from denied to granted, ad tags can set and read those identifiers again.
- Added in
- Original version (GDPR-era Consent Mode)
- Governs
- Cookie and identifier writes for advertising
Key pointCheck it first: almost every Ads measurement issue traces back here.
Analytics_storage
Controls whether analytics cookies can be written and read. When it flips to granted, GA4 writes client IDs and session cookies normally instead of relying on cookieless pings.
- Added in
- Original version
- Governs
- Analytics cookie read and write
Key pointDenied here means GA4 falls back to modeling and pings.
Ad_user_data
Controls whether user data may be sent to Google for advertising purposes. When it flips to granted, conversion and audience data can flow to Google Ads for measurement and bidding.
- Added in
- November 2023
- Governs
- Sending user data to Google for ads
Key pointWithout it, Ads loses the data it uses for bidding and remarketing.
Ad_personalization
Controls whether ads can be personalized for that user. When it flips to granted, remarketing lists and personalized targeting become available for that visitor.
- Added in
- November 2023
- Governs
- Ad personalization and remarketing
Key pointDenied does not stop ads; it stops tailoring them to the individual.
Going further
Understand the two new signals
Concept
Why the two new signals matter more than the old two
ad_user_data and ad_personalization were split out because sending data and personalizing ads are legally distinct from storing cookies.
The original two signals answered one question: can this device be identified and measured? ad_storage covered advertising identifiers, analytics_storage covered analytics cookies. Regulators increasingly treated data transmission and ad personalization as separate purposes, so Google split them into their own switches in November 2023.
This matters for advertisers who previously relied on ad_storage alone. A visitor can now allow advertising cookies but refuse having data sent to Google, or allow data transmission but refuse personalization. Treating ad_storage as a proxy for the other two is the most common misconfiguration in Consent Mode v2.
For a small team, the practical consequence is that your banner must map each purpose to the correct signal. A single 'accept ads' toggle that only sets ad_storage leaves ad_user_data and ad_personalization denied by default, which quietly degrades Ads bidding.
- Banner-to-signal mappingConfirm each banner purpose sets its matching signal, not just ad_storage.
- Default valuesVerify all four signals are declared in the default state before any interaction.
- Ads-only setupsSites with no GA4 still need analytics_storage if any analytics tag runs.
- Regional scopeCheck whether the default state is set globally or only for EEA visitors.
02 / 10
What is the difference between the default and update consent state?
The default state applies before the visitor makes a choice; the update state applies the moment their choice changes, and the two must be verified separately.
No consent values exist until a page explicitly pushes a default state, the set of yes/no values the page assumes before the visitor makes any choice. That push belongs on every page load, first thing, with consent signals denied for any region where your banner is shown.
The update state is the new set of values sent when the visitor clicks the banner; only then does that session change. Test the two separately, because a correct default can hide a broken update, and a working update can mask a default that never fired.
Default pushed on every page load
Confirm the default consent command runs before any tag and on every template, not just the homepage. A single page that loads tags first will send data before consent.
Region scoping matches your banner
Scope the default to the regions where your banner actually appears. If your banner shows only to EEA visitors, your default should name those regions; a global default would deny consent for visitors who never saw a banner.
Update wired to the banner callback
Check that the update command fires from the banner's accept or reject callback, not on a timer. Verify it runs once per choice, since a repeated update can overwrite a later decision.
03 / 10
Should I use basic or advanced consent mode?
Basic blocks Google tags entirely until consent; advanced loads them with default settings and lets them adjust to the current state, which is what unlocks modeling.
Basic mode blocks Google tags until consent is granted, so it is simple to reason about but you lose all data from visitors who never consent. Advanced mode loads tags with default settings and lets them adjust to the current consent state, keeping a cookieless ping flowing when consent is denied. That ping feeds conversion modeling, Google's estimation of conversions it cannot observe directly.
The trade-off is complexity and verification: with advanced mode you must confirm tags still behave correctly under denied consent, since they run from the first pageview. Choose based on who maintains the setup and how much unobserved conversion data you can afford to lose.
Basic mode
- Blocks Google tags entirely until consent is granted
- Easier to explain and audit because no tag fires before consent
- No cookieless ping, so no conversion modeling for unconsented visitors
- You lose all measurement from visitors who never consent
Fits small in-house teams that prioritize simplicity and can tolerate missing data from non-consenting visitors.
Advanced mode
- Keeps a cookieless ping flowing when consent is denied
- Feeds conversion modeling to estimate unobserved conversions
- Tags load with default settings from the first pageview
- Requires verifying that tags still behave correctly under denied consent
Fits agencies and in-house teams that need modeled conversions and can maintain and verify the more complex setup.
04 / 10
How do I confirm the consent state in Tag Assistant and GTM Preview?
Open Tag Assistant, load the page, read the consent state it reports, then interact with the banner and re-check that the state changed.
Tag Assistant reports the consent state Google sees for the page you loaded. A CMP, the Consent Management Platform that collects and stores visitor choices, talks to Google through the consent API, and Tag Assistant reads what arrives.
GTM Preview shows the same moment tag by tag: firing order, which tags ran, and which were blocked by consent. If the reported state still reads denied one second after you click Accept, the update is not reaching your tags, check the banner callback next.
-
Open Tag Assistant
Connect Tag Assistant to the domain, load the page in a clean tab, and let the container finish loading.
-
Read the default state
Before touching the banner, read the consent state Tag Assistant reports. This is the state applied before any visitor interaction.
-
Interact with the banner
Click Accept (or Reject) once, and wait for the CMP to confirm the change in its own interface.
-
Re-check the state
Return to Tag Assistant and read the consent state again. If it still reads denied after Accept, the update is not reaching your tags.
-
Confirm tag behavior in GTM Preview
Switch to Preview, repeat the same sequence, and watch which tags fire or are blocked after the consent change.
05 / 10
How do I read the gcs parameter in a network request?
Open the browser network tab, filter for requests to Google endpoints, and read the gcs value, which encodes the ad_storage and analytics_storage state.
The gcs parameter, a short code attached to Google requests that encodes the current consent state, appears on hits sent to Google endpoints from your tags. Open DevTools, filter the request list to those domains, and check whether the value matches what you selected before and after a consent change.
That comparison proves the state set in Tag Assistant is transmitted on the wire, not merely stored locally. If the code does not flip when consent is granted or denied, the tag is not propagating consent.
-
Open the network panel
In Chrome DevTools, go to the Network tab and reload the page with your consent banner reset.
-
Filter for Google requests
Type a Google collection endpoint in the filter box so only tag hits remain in the list.
-
Read the gcs value
Click a request, open the query string parameters, and find the gcs key.
-
Compare before and after
Grant consent in the banner, trigger a new hit, and confirm the gcs value changed.
Going further
Decode the gcs value
Understanding the calculation
Decoding the gcs value
gcs carries the two original storage signals; the two signals added in v2 travel in a separate parameter, gcd.
The gcs value starts with G1, followed by one digit for ad_storage and one for analytics_storage: 1 means granted, 0 means denied.
So G100 means both storage signals are denied, G111 means both are granted, and G101 means ad_storage denied with analytics_storage granted. A value that never moves away from G100 after someone accepts the banner points to an update that fires too late or not at all.
ad_user_data and ad_personalization, the two v2 signals, are not in gcs. They are sent in the gcd parameter, whose encoding is longer; read it in Tag Assistant's consent view rather than decoding it by hand.
- Before consentCapture one hit with the banner still open and note the gcs value (G100 under a deny-by-default setup).
- After consentAccept the banner, trigger a new hit, and confirm the value moved to G111.
- Partial consentAccept analytics only and check that the value reads G101.
- v2 signalsCheck ad_user_data and ad_personalization in Tag Assistant's consent view, since gcs does not carry them.
06 / 10
What actually happens to GA4 and Google Ads data when consent is denied?
In advanced mode, tags still load and send cookieless pings that Google can use for modeling; in basic mode, tags are blocked and no data flows until consent is granted.
In advanced mode, a denied analytics_storage signal does not blank out GA4. Tags still fire and send cookieless pings containing no identifiers; Google can use those pings to feed conversion modeling, its estimation of conversions it cannot observe directly when consent is missing. So you lose identifiers, not the event itself, expect lower counts, with modeling filling part of the gap.
When the ad signals (ad_user_data, ad_personalization) are denied, Google Ads keeps the cookieless ping but cannot build remarketing audiences or serve personalized ads from that traffic. In basic mode, nothing fires at all until consent is granted, so there is no estimate to model. Hits collected while consent was denied on the same page do not need to be resent once consent is granted.
GA4 with analytics_storage denied
In advanced mode, tags load and send cookieless pings with no user identifiers. GA4 records fewer events and uses conversion modeling to estimate part of what it cannot observe directly.
- Signals sent
- Cookieless pings only
- Mode that applies
- Advanced
Key pointCheck that the ping is present and that modeling is enabled in your GA4 property, then treat the modeled share as an estimate, not a count.
Google Ads with ad signals denied
ad_user_data and ad_personalization set to denied stop remarketing lists and personalized ad delivery from that traffic. The cookieless ping still supports aggregate reporting and conversion modeling.
- Signals sent
- Cookieless pings only
- Remarketing
- Not available
Key pointVerify both ad signals in Tag Assistant, because a single denied flag is enough to switch off personalization.
Consent granted later on the same page
If a visitor accepts after a denied hit, the next updates are sent with the new state and tags resume normal behavior. You do not need to resend the earlier denied hits.
- Resend needed
- No
- Cookie writes
- Allowed from the update
Key pointConfirm the state change appears in the network request once consent is granted, then leave the earlier pings as they are.
07 / 10
Why does my site say consent not detected?
Most "consent not detected" cases come from a state that is never pushed, an update that fires before the default, or a banner that reports back after the tag has already run.
A banner that loads and collects a choice can still leave your Google tags reading the default state, the consent values a page assumes before the visitor has made a choice. If that state is never pushed on a given page load, or if an update runs before the default, Google tags see no consent at all. The CMP (Consent Management Platform, the tool that collects and stores visitor choices) can look healthy while the values never leave the banner.
Third-party scripts loaded outside your tag manager can fire before consent is set, overwriting or masking the real state. Use the troubleshooting table below to map the symptom to the cause and the fix, then confirm the state in Tag Assistant before you move on.
Symptom, cause, and fix for consent detection failures
| Symptom | Likely cause | Fix to apply |
|---|---|---|
| No consent state reported at all | Default state is never pushed on that page load | Push the default state before any Google tag runs on every page, including landing pages |
| State stays denied after clicking Accept | Update callback runs too early or not at all | Fire the update after the banner resolves, inside the CMP callback |
| Tag fires on deny | Tag loaded outside your tag manager before consent is set | Move the tag into GTM and bind it to the consent signal, or remove it |
Diagnostic patterns for Consent Mode v2; fixed causes, not forecasts.
Going further
See the failure patterns
Special case
The failure patterns that hide behind a working banner
The less obvious failures hide behind a banner that looks fine but never delivers its state to Google tags.
A working banner can still fail silently. The CMP might store the choice in its own cookie and update the UI, but the consent state never reaches the data layer or the Google tag. This often happens when the CMP integration is configured for display only or when the update callback is registered on a different event than the one the tag listens to.
Third-party scripts loaded outside your tag manager are another blind spot. A script injected by a plugin or a direct script tag can fire before the CMP has set any state, sending a hit with no consent information. Google tags then read whatever default was in place, often denied, and you lose the data without any visible error.
In advanced mode, the default state must be pushed on every page load, including pages where the CMP has not yet initialized. If the default is missing, Google tags may not send any hits at all, or they may send them with an empty gcs parameter. Checking the gcs parameter in a network request is the quickest way to confirm what the tag actually saw.
- Default state on every pageConfirm the default consent state is pushed before any Google tag fires, on every page load, including landing pages where the CMP may not have initialized yet.
- Update callback timingVerify the update callback runs after the banner resolves. An early callback leaves Google tags reading stale values.
- Tags outside the tag managerAudit for third-party scripts loaded outside GTM that can fire before consent is set and mask the real state.
- gcs parameter in network requestsRead the gcs parameter in a network request to see exactly what consent signal the tag sent.
08 / 10
What should I monitor after I deploy Consent Mode v2?
Set a weekly routine: check the consent state on a live page, watch GA4 traffic and Google Ads conversions for a defined grace period, and only then decide whether measurement is broken.
Give the change a grace period before you call it lost data. Pick a fixed window, often two weeks, and only treat a dip as broken measurement after that window closes.
Each week, check the consent state on a live page and watch the ratio of consented to denied sessions, not just the total. Set the owner of that check in advance, and agree when to loop in your CMP vendor (the tool that collects and stores visitor consent choices) or implementation partner.
Weekly consent state check
Open a live page and confirm the default state shows before interaction and the update state shows after a choice. If either is missing, escalate to the tag owner that week.
Consented vs denied session ratio
Track both counts separately. A falling total with a stable ratio points to a tracking issue; a falling ratio points to a consent or banner issue.
Grace period on reported data
Name the end date of your window, usually around two weeks. Do not declare lost measurement before it, because modeled conversions fill gaps gradually.
Named owner and escalation path
Write down who runs the check each week and who to contact at your CMP vendor or implementation partner. A rule with no owner tends to be skipped.
09 / 10
Frequently asked questions
10 / 10




