Google Ads Consent Mode v2: What the June 2026 Change Actually Means for Your Conversion Numbers

Open a Google Ads account that runs any meaningful volume of traffic from the EU, the UK, or the wider EEA, and there’s a decent chance you’ll find a small, unlabeled discrepancy between what your analytics platform says converted and what Google Ads is reporting. Not a dramatic one. A conversion count that’s slightly higher than it should be, or slightly lower, with no campaign change, no tracking code edit, and no obvious explanation in the change history. Support forums are full of advertisers asking a version of the same question: “My conversions didn’t change, my traffic didn’t change, so why did my numbers move?”

Most of the time, the answer isn’t a bug. It’s consent mode doing exactly what it’s designed to do — filling gaps in your data with statistical estimates, because a portion of your visitors declined tracking and Google is modeling what probably happened to them instead of just leaving those numbers blank. That’s a reasonable thing for the system to do. The problem is that most advertisers never explicitly agreed to it, don’t know it’s happening, and have no idea how to tell a modeled conversion from a real one when they’re staring at the same “Conversions” column.

This gets more pressing in 2026 because Google has been tightening how this whole system works, on a schedule that isn’t optional for anyone advertising to EEA or UK users. There’s a hard deadline landing this year that changes how much of your ad data flows to Google Ads if your consent signals aren’t wired up correctly. If you manage an account that touches European traffic — even a small slice of it — this is worth twenty minutes of your time before it costs you a much bigger chunk of your budget.

What Consent Mode Actually Is, in Plain Terms

Consent Mode is Google’s mechanism for reconciling two things that are otherwise in direct conflict: privacy regulation that requires a user’s explicit permission before you track them, and an advertising platform that runs on tracking data. Instead of a binary “tag fires or it doesn’t,” consent mode lets your site tell Google, on every single page load, what the visitor actually agreed to. Four signals do the talking:

• ad_storage — whether cookies related to advertising (like the click ID used for conversion attribution) can be stored on the user’s device.
• analytics_storage — whether analytics cookies, like those used by GA4, can be stored.
• ad_user_data — whether user data can be sent to Google for advertising purposes at all. This is the one that governs whether Google can use the data for ad measurement in the first place.
• ad_personalization — whether that data can be used for personalized advertising and remarketing specifically.

The first two signals existed in the original version of consent mode. ad_user_data and ad_personalization were added in Consent Mode v2, which Google introduced to align with the EU’s Digital Markets Act. That distinction matters because a lot of sites still running an old v1 implementation — correct as of 2023 — are quietly non-compliant now, and in some cases have been losing ad measurement capability for EEA and UK traffic without anyone on the team noticing, because nothing throws an error. The tag just stops sending what it used to send.

When a visitor declines consent, the relevant Google tags don’t fire in the way they normally would. What happens next depends on which of two implementation modes you’re running, and this is where a lot of the confusion in Google Ads reporting actually originates.

Basic vs. Advanced Consent Mode: The Distinction That Decides How Much Data You Keep

Basic consent mode is the simpler of the two, and also the more restrictive. Under basic mode, Google’s tags are blocked outright until a visitor grants consent. If they decline, nothing is sent for that visit — no cookieless signal, no ping, nothing for Google’s models to work with beyond the fact that a page loaded. This is compliant, and it’s easy to implement, but it means every visitor who says no becomes a complete blind spot in your reporting.

Advanced consent mode behaves differently. Tags still load, but when consent is denied, they send anonymous, cookie-less “pings” instead of full tracking data — a signal that a page view or a conversion event happened, without any of the identifying information that would normally accompany it. No cookie is set, no personal data is transmitted, but Google now has a data point it can feed into a statistical model. That model looks at the behavior of visitors who did consent, in the same country, on the same domain, around the same time, and uses those patterns to estimate what happened with the visitors who didn’t.

This is where “modeled conversions” come from. They’re not fabricated numbers pulled from nowhere — they’re a genuine statistical estimate built from your own consented traffic. But they are estimates, not measured events, and the accuracy of that estimate depends heavily on how much consented traffic you actually have to model from. An account running high volume in a country with decent consent rates can get a reasonably tight estimate. A smaller account, or one operating in a market where consent refusal rates run high — and several EU markets do — may not have enough of a consented sample for the modeling to produce anything reliable, which is part of why some advertisers see conversion numbers that don’t fully reconcile with what their CRM or backend system is recording.

The practical decision here isn’t really “which mode is better.” It’s a tradeoff every business running EEA or UK traffic has to make deliberately: basic mode is more conservative and leaves more of your funnel invisible, advanced mode recovers more signal but relies on modeling assumptions you don’t fully control. What’s not acceptable is not knowing which one your site is actually running, which happens more often than you’d expect when a consent management platform (CMP) was installed by a developer or agency that’s no longer involved with the account.

How Conversion Modeling Actually Behaves Inside Your Reports

Once advanced consent mode is active and Google has enough data to model against, modeled conversions start appearing in your standard “Conversions” column in Google Ads — they’re not broken out into a separate line by default, which is exactly why so many advertisers don’t realize part of their reported number is an estimate rather than a count. Modeling requires a minimum volume of consented and non-consented traffic before it activates at all; below that threshold, Google simply won’t model, and you’ll see the gap show up as fewer recorded conversions instead. Above the threshold, modeled conversions get folded into every downstream report that reads from the conversions column — campaign performance, Smart Bidding, attribution reporting, all of it.

That last point is the one worth sitting with. Target CPA and Target ROAS bidding strategies optimize toward the conversion data Google Ads gives them, and they don’t distinguish between a conversion that was directly observed and one that was modeled from a statistical inference. If you want more background on how those bidding strategies actually respond to the conversion signal they’re fed, our comparison of Target CPA, Target ROAS, and Maximize Conversions goes into how sensitive each strategy is to the quality of the input data. A bidding algorithm chasing a target built partly on modeled numbers is, in effect, optimizing toward its own estimate of itself — usually fine, occasionally a feedback loop worth watching if your modeled share is high.

This also connects directly to attribution. Google consolidated its attribution models down to two options — last click and data-driven attribution — and data-driven attribution in particular leans on a large, clean dataset to build its credit-assignment model. If a meaningful share of your conversions are modeled rather than observed, the attribution model is training on a mix of real behavior and inferred behavior, which is worth knowing before you draw hard conclusions about which keyword or campaign actually deserves the credit. We covered the mechanics of that shift in more detail in our piece on the 2026 attribution model changes, and it’s worth reading alongside this if you’re trying to reconcile attribution behavior with consent-driven data gaps.

The June 15, 2026 Change: Why Google Signals Retiring Actually Matters

Here’s the part of this that has a real deadline attached to it rather than being a general best-practice recommendation. For years, a lot of advertisers who hadn’t fully wired up consent mode were still getting a partial safety net through Google Signals — the mechanism that let Google Analytics data cross over and backfill some Google Ads measurement even when the ads-specific consent signal wasn’t cleanly configured. It wasn’t a designed feature so much as a side effect of how the two products’ data pipelines were connected, but plenty of accounts came to depend on it without realizing that’s what was happening.

On June 15, 2026, that cross-influence goes away. Google is narrowing the role of ad_storage so that it becomes the sole governing signal for advertising data in Google Ads, cutting the link that let Google Analytics settings quietly compensate for gaps on the ads side. If your ad_storage configuration is correct and your CMP is passing consent signals properly, this change shouldn’t be visible to you at all — you were never actually relying on the fallback. But if you were leaning on Google Signals to paper over an incomplete or misconfigured consent mode setup, this is the point where that stops working, and the drop in reported conversions won’t come with a warning banner explaining why.

This sits on top of an enforcement timeline that’s already been running for a while, not a brand-new requirement appearing out of nowhere. Consent Mode v2 became mandatory for EEA and UK data collection back in March 2024. Google began actively enforcing that requirement in mid-2025, and accounts that hadn’t implemented it correctly started losing access to personalized advertising, remarketing, and reliable conversion measurement for that traffic — quietly, without a grace period announcement most people saw. The June 2026 change is best understood as the next link in that same chain: Google steadily removing the workarounds and fallbacks that let incomplete implementations keep functioning. Each step narrows the gap between “technically has consent mode installed” and “has it installed correctly enough that it actually protects your measurement.”

What this means practically: if any part of your team’s mental model is still “we set up a cookie banner a couple of years ago, so we’re covered,” that’s the exact assumption worth re-checking this year, specifically before the June date.

Where Consent Mode Setups Actually Go Wrong

Most broken consent mode implementations aren’t broken because someone skipped it entirely. They’re broken because of a handful of specific, fixable gaps that tend to repeat across accounts:

• The CMP and the tag aren’t actually talking to each other. A cookie banner can be fully compliant from a legal-consent-string perspective and still fail to update the gtag or Google Tag Manager consent state, because the integration between the CMP and the tagging layer was never wired up, or broke silently after a platform update.
• No default consent state is set before the CMP loads. Google expects a default state (usually “denied” for EEA/UK traffic) to be declared before the consent banner even renders, so that nothing fires in an undefined state during the brief window before the user makes a choice. Sites that skip this default end up either over-firing tags before consent is captured, or under-firing them everywhere by mistake.
• Region targeting isn’t configured, so the same default applies globally. Consent mode supports geo-targeted defaults, letting you apply stricter defaults only to EEA/UK visitors while leaving other regions unaffected. Accounts that apply one blanket setting everywhere often end up either needlessly restrictive on traffic that didn’t require it, or non-compliant on the traffic that did.
• Server-side tagging setups that don’t forward consent state correctly. Server-side GTM is increasingly common, and it’s easy to get the event data flowing while missing that the consent signal itself needs to be forwarded alongside it, not just referenced client-side.
• Nobody has looked at it since the initial setup. Consent banners get redesigned, CMPs get swapped for cheaper vendors, sites get rebuilt on new platforms — and consent mode wiring is exactly the kind of infrastructure that doesn’t announce itself when it breaks. It just quietly stops sending signals, and the first symptom anyone notices is a conversion count that looks off.

None of these require a developer to catch, at least at the diagnostic stage. Google’s own Tag Assistant, and the consent preview mode built into Google Tag Manager, will show you in real time whether ad_storage, ad_user_data, and ad_personalization actually change state when you interact with your own cookie banner. If you click “reject all” and the debug view still shows those signals as granted, or if they never update at all, you’ve found your problem in about five minutes.

Enhanced Conversions as the Other Half of This Story

Consent mode and Enhanced Conversions solve related but distinct problems, and it’s worth being clear about which is which. Consent mode governs whether Google can collect signal at all, given what the user agreed to. Enhanced Conversions improves the accuracy of the conversion events you’re already allowed to send, by matching hashed first-party data — an email address or phone number a customer gave you directly — against a Google-side identity, closing gaps caused by ad blockers, cross-device journeys, or browsers that limit third-party cookies entirely.

They work best together. A visitor who declines advertising cookies but completes a purchase and hands over their email at checkout can still, depending on your setup, contribute a matched, first-party-verified conversion through Enhanced Conversions — recovering signal that consent mode’s cookieless pings alone wouldn’t fully capture. We’ve written before about how often Enhanced Conversions setups look correct in the interface but are silently failing to match anything, which is worth checking alongside your consent mode audit rather than treating them as two separate projects. That earlier piece, on what Enhanced Conversions actually fix and why they’re often silently broken, walks through the specific things to check in the diagnostics tab.

The same first-party data discipline shows up again if you’re building remarketing or Customer Match audiences from the same customer list. Both rely on the same hashed-match infrastructure, and both are affected by the same consent requirements — a user who declined ad personalization can’t legally be added to a personalized remarketing list even if you have their email address from a purchase. We go into how that list-building process actually works, and where the 2026 rules changed it, in our guide to Customer Match, which is worth reading alongside this if first-party audiences are a meaningful part of your account strategy.

If you’re auditing conversion tracking end to end — which is the right instinct any time your reported numbers don’t match what your CRM or backend order system is recording — consent mode is one input into a bigger picture that also includes tag firing order, deduplication between GA4 and Google Ads, and value passback accuracy. We laid out a fuller audit process in our guide to auditing Google Ads conversion tracking, and consent mode is worth adding as an explicit checklist item there if it isn’t already, particularly for any account with EU or UK traffic mixed in.

What to Actually Check, in Order

If you manage an account with any EEA or UK traffic and you haven’t looked at this since it was first set up, here’s a reasonable order to work through it in:

• Confirm which consent mode version is live. If your implementation predates 2024 and hasn’t been touched, assume it’s v1 until proven otherwise, and check specifically for ad_user_data and ad_personalization in your tag configuration — their absence is the clearest sign you’re still on the old version.
• Decide, deliberately, between basic and advanced mode. Don’t leave this as an accident of whichever CMP default got installed. If measurement quality on EU traffic matters to the business, advanced mode with modeling is usually the better tradeoff, but the decision should be made and documented, not defaulted into.
• Test the actual consent flow, not just the presence of a banner. Use Tag Assistant or GTM’s preview mode, walk through accepting and then separately rejecting consent, and confirm the signal states actually change in the debug output both times.
• Check your default consent state and region targeting. Confirm a “denied” default is set before the CMP loads for EEA/UK visitors specifically, and that you’re not applying an overly broad or overly narrow region rule.
• If you’re on server-side tagging, verify consent is forwarded, not just events. This is the gap that catches out teams who migrated to server-side GTM for performance reasons and didn’t revisit the consent wiring afterward.
• Look at your reported conversion volume for a “modeled” pattern. A conversion count that fluctuates in ways that don’t track with your actual traffic or sales volume, especially concentrated in EU-heavy campaigns, is worth cross-referencing against your CRM numbers for the same period.
• Put June 15, 2026 on the calendar specifically. Even if everything above checks out today, re-verify it again in the days after that date, since it’s the point where any latent misconfiguration that was previously masked by the Google Signals fallback becomes visible in your numbers.

It’s also worth budgeting real time for this rather than treating it as a five-minute toggle. Coordinating a consent mode fix usually means pulling in whoever owns the CMP vendor relationship, whoever has access to the Google Tag Manager container, and whoever can verify the change against the live site once it’s pushed — three different people at a lot of companies, and sometimes three different agencies. Scoping that out in advance, rather than discovering the coordination problem in the middle of a June 2026 conversion drop, is the difference between a routine fix and a scramble.

None of this is a one-off task you complete and forget. Consent management platforms get updated by vendors, sites get redesigned, tagging gets migrated to new platforms, and any of those events can quietly break a signal that was working fine the week before. That’s true of a lot of the infrastructure sitting underneath a Google Ads account — it doesn’t fail loudly, it just stops doing its job while every dashboard keeps rendering normally. This is part of why we built continuous, daily monitoring into Growera’s Google Ads management rather than treating audits as something you do quarterly: a consent signal that stops firing, or a conversion count that quietly drifts against your traffic and revenue trends, is exactly the kind of change that’s easy to catch on day one and expensive to catch three months later, once Smart Bidding has already adjusted to the wrong numbers.

What This Means If You Don’t Run EU Traffic

It’s worth being direct about who this actually affects. If your account runs exclusively on US or other non-EEA, non-UK traffic, none of the consent mode requirements above are legally mandatory for you today, and the June 2026 change won’t touch your reporting. But a few things make this worth reading anyway rather than skipping. Privacy regulation in this space tends to expand rather than contract — several US states have their own data privacy laws with their own consent implications, and platform-level changes like third-party cookie deprecation are pushing every advertiser toward first-party-data-dependent measurement regardless of jurisdiction. Enhanced Conversions, in particular, is worth implementing everywhere, not just where consent mode is legally required, because the underlying problem it solves — measurement gaps from cookie loss, ad blockers, and cross-device behavior — isn’t specific to Europe.

If your business does have any EU or UK customers at all — a SaaS product with international signups, an ecommerce store that ships internationally, an agency managing accounts for clients with EU operations — even a small percentage of that traffic is enough to trigger the compliance requirement for the account as a whole. It’s a common blind spot for B2B and SaaS advertisers in particular, who sometimes assume consent mode is a concern for European retailers rather than for accounts with a global customer base and only a modest EU segment. Our page on Google Ads management for B2B and SaaS businesses covers the wider measurement setup that tends to come up for that kind of account, consent mode included.

The Honest Limits of What Modeling Can Fix

It’s worth resisting the temptation to treat advanced consent mode and conversion modeling as a full replacement for the tracking you lose when a user declines consent. It isn’t. It’s a partial, statistical compensation, and its accuracy is bounded by how much consented data you actually have to model from. A large ecommerce account running significant volume across multiple EU markets will get a reasonably solid model. A smaller account, or one concentrated in a market with a high consent-refusal rate, may see modeled conversions that carry a wide margin of error, or may not clear the volume threshold for modeling to activate at all in certain segments — meaning those conversions simply don’t show up anywhere, modeled or otherwise.

There’s also a reporting-transparency issue worth being upfront about: Google Ads doesn’t make it particularly easy, in the standard interface, to see what share of your reported conversions in a given period were modeled versus directly observed. That visibility exists in more detail in Google Analytics’ modeling reports, but the Google Ads conversions column itself blends the two without a default breakdown. If you’re making a significant budget decision based on a conversion trend, and a meaningful share of your EU traffic is running through consent mode modeling, it’s worth treating that number with slightly more caution than a fully observed conversion count from a market where consent mode isn’t triggering modeling at all.

None of this is a reason to avoid advanced consent mode — the alternative, basic mode, gives you less data, not more accurate data. It’s a reason to understand what you’re looking at, rather than treating every number in the conversions column as equally solid ground to build a bidding strategy or a budget decision on. When a number in a report can be either a directly observed event or a statistical inference, and the interface doesn’t tell you which, a little healthy skepticism about month-to-month swings in EU-heavy segments is just good practice, not paranoia.

Summary

Consent Mode v2 isn’t new, but the way Google is enforcing it keeps changing, and June 15, 2026 is the next real deadline on that timeline — the point where Google Signals stops quietly backfilling gaps in an incomplete Google Ads consent setup. If your implementation is correct, this changes nothing for you. If it isn’t, this is the point where that stops being invisible.

The practical takeaway is simple even if the underlying mechanics aren’t: know whether you’re running basic or advanced consent mode and make sure that was a deliberate choice, verify with Tag Assistant or GTM preview mode that your consent signals actually change state when a visitor makes a choice, check that a default “denied” state is set for EEA/UK traffic before your CMP loads, and don’t assume a cookie banner installed a couple of years ago is still doing its job correctly today. Pair that with a working Enhanced Conversions setup so you’re not relying on modeling alone to recover signal, and revisit the whole thing again shortly after June 15 to make sure nothing that was previously masked by the old fallback has come loose. None of this is complicated once you know where to look. It’s just easy to leave unchecked precisely because, unlike a paused campaign or a rejected ad, a broken consent signal never tells you it’s broken. Your reports just quietly become a little less true, month after month, until someone finally goes looking.