What Consent Mode v2 Does to Your Client Reports

What consent mode v2 does to your client reports

A client emails on a Tuesday asking why their traffic fell off a cliff in one month and never came back. You check the campaigns. Spend is steady, rankings are fine, the phone is still ringing. Nothing is broken. What changed is that their developer finally wired up a cookie banner properly, and GA4 stopped counting the people who said no.

That is the consent mode v2 reporting impact in one sentence: the numbers in your report stopped describing everyone who visited and started describing everyone who consented, plus a modelled estimate of the rest. The marketing did not get worse. The measurement got narrower and more honest at the same time.

This post is the explainer we wish existed when we started fielding these questions. What consent mode v2 is, exactly which metrics it touches and which it leaves alone, how modelled conversions actually work (including the thresholds most small clients never hit), and the plain-English wording you can put in front of a client without sounding like you are making excuses. We also pulled numbers from our own Google Analytics integration data to show how often this comes up in real client accounts.

Last updated: October 2026

Key takeaways

• Consent mode v2 is a signal layer, not a tracking tool. It tells Google tags whether a visitor consented to analytics storage, ad storage, ad user data and ad personalisation, and the tags change behaviour accordingly.
• Google documents two implementations. Basic consent mode blocks Google tags until a visitor accepts, so denied visitors are never seen at all. Advanced consent mode loads tags immediately in a denied state and sends cookieless pings, which is what makes modelling possible.
• Modelled data is not free. Google Analytics Help states behavioural modelling needs at least 1,000 events per day with analytics_storage denied for 7 days, plus at least 1,000 daily users with consent granted on 7 of the previous 28 days. Most small-business clients never clear that bar.
• The gap is unevenly spread. Sessions, users and conversions in GA4 are affected. Google Search Console and Google Business Profile data are not, because they come from Google's own logs rather than a browser cookie.
• In our own book of business, Google Analytics is the single most-connected reporting source (30 of 43 connected client accounts, roughly 7 in 10), so consent changes hit the exact metric set most agency reports are built on.

On this page

• What is consent mode v2, in plain English?
• What is the consent mode v2 reporting impact on an agency report?
• Modelled conversions explained
• Basic vs advanced consent mode and why your report looks different
• Which reporting sources cookie consent cannot touch
• What our own client data shows
• How to explain consent mode data gaps to a client
• A reporting checklist for consent mode v2
• FAQs

What is consent mode v2, in plain English?

Consent mode v2 is Google's way of receiving a visitor's cookie choice as a machine-readable signal, so Google tags can behave differently for people who accepted and people who did not. Google Ads Help describes it directly: "Consent mode lets you communicate your users' cookie or app identifier consent status to Google. Tags adjust their behavior and respect users' choices."

The "v2" part added two parameters to the original pair. Version one carried analytics_storage and ad_storage. Version two added ad_user_data and ad_personalization, which is what Google requires before a visitor's data can be used for audience building and personalised advertising in the European Economic Area and the UK.

Two bits of insider vocabulary worth defining for your clients, because they will hear both:

Cookieless pings. When consent is denied under advanced implementation, the tag still fires, but it sends an anonymous signal with no cookies and no persistent identifier. Google uses the metadata (timestamp, user agent, referrer, consent state) as raw material for modelling. It does not identify the visitor.

Modelled data. An estimate produced by Google's machine learning, filling the measurement gap between consented visitors and everyone else. It appears inside normal GA4 metrics without a visible label, which is exactly why it causes arguments in client meetings.

Here is the part agencies underrate: consent mode is not a cookie banner. The banner collects the choice, a consent management platform stores and republishes it, and consent mode is the handshake between that platform and Google's tags. Three separate pieces, three separate places it can be set up wrong, and only one of them is something your client's developer thinks of as "analytics".

What is the consent mode v2 reporting impact on an agency report?

The consent mode v2 reporting impact is concentrated in browser-measured metrics and absent from server-measured ones. That single distinction explains almost every discrepancy a client will point at.

If a metric needs a cookie in the visitor's browser to exist, consent can suppress it. If a metric is counted on Google's side of the wire, consent never enters the picture. Sessions in GA4 need a cookie. Impressions in Search Console do not.

The asymmetry in that table is the whole problem. Your media numbers stay intact while your outcome numbers shrink, so every cost-per-conversion figure in the report moves in the wrong direction at once. A client reading quickly will conclude the campaigns got more expensive. They did not. The denominator changed.

The second-order effect is channel attribution. When a tag cannot read a cookie, GA4 loses the stored campaign reference, and that traffic tends to resolve as Direct or sit in Unassigned. If you have ever had to explain a sudden Direct spike to a client who did not change a thing about their marketing, that is usually the mechanism at work.

Modelled conversions explained

Modelled conversions are Google's statistical estimate of outcomes it could not observe directly, inferred from the behaviour of consented visitors with similar characteristics. They are not a guess pulled from nowhere, and they are not a measurement either. They sit in between, and that is the honest framing to give a client.

Mechanically, Google trains on the consented population (which device, which channel, which time of day, which pages, how often that combination converts), then applies the learned pattern to the cookieless pings from the denied population. Google Ads Help draws a sharp line between the two implementations here: advanced consent mode supports an "advertiser-specific model (more detailed modeling)" while basic consent mode only qualifies for a "general model (less detailed modeling)".

Then come the thresholds, and this is where most agency clients fall out of the story entirely. Google Analytics Help publishes the requirements for behavioural modelling:

Read those two volume lines against a typical local client. A dentist, a law firm, a trades business, a regional retailer: a thousand daily users is a number most of them will not see this year. For those accounts the modelling never switches on, so the data gap stays a hole rather than becoming an estimate. Telling a client "Google will fill it in" when their site does 300 sessions a week is a promise you cannot keep.

It also matters where modelled data does and does not appear. Google Analytics Help states modelling feeds standard reports (affecting users, sessions and new users), and explicitly lists where it is not applied: audiences, user explorer, cohort and user lifetime explorations, segments with a sequence, retention reports, predictive metrics, and data export including BigQuery export.

That last exclusion is the one that ambushes agencies. A number in a standard GA4 report can legitimately disagree with the same number pulled from BigQuery, because one includes modelled data and the other never does. If you build client reports off a warehouse export, you are reporting the observed-only view while your client is looking at the modelled view in the GA4 interface. Our breakdown of GA4 data sampling and thresholding covers the other quiet reasons two GA4 numbers refuse to match.

Basic vs advanced consent mode and why your report looks different

Basic and advanced consent mode produce visibly different reports from the same cookie banner and the same traffic. Knowing which one a client runs is the first diagnostic question, not a technical footnote.

Neither column is the universally right answer, and an agency should not pretend to be the one making that call. Advanced recovers more measurement. Basic sends nothing before consent, which some European regulators and privacy counsel regard as the safer reading of ePrivacy rules, since advanced transmits a ping before the visitor has chosen. That is a legal decision for your client and their adviser. Your job is to know which mode is live, because it determines whether the gap in the report is a hole or an estimate.

We built ReportsMate email-first after years around agency reporting, and the pattern we kept seeing is that nobody loses a client over a consent-driven dip. They lose them over an unexplained consent-driven dip that sat in a dashboard for three months with no commentary attached. The explanation has to travel with the numbers.

Which reporting sources cookie consent cannot touch

Search Console, Google Business Profile and in-platform ad delivery metrics are unaffected by cookie consent, because none of them depend on a cookie being written in the visitor's browser. They are counted on Google's and Meta's own infrastructure.

That makes them the stable spine of a client report in a consent-restricted world:

• Google Search Console. Clicks, impressions, average position and query data come from Google's search logs. A visitor who rejects every cookie on arrival still generated a Search Console click. For organic reporting this is the number that does not move, and the difference between it and GA4 organic sessions is now structural rather than a bug. We unpacked that specific mismatch in Search Console vs GA4 organic traffic.
• Google Business Profile. Search and maps impressions, website clicks, calls and direction requests are counted on Google surfaces before the visitor reaches the client's site at all. For local clients, GBP is often the most complete demand signal you have left.
• Ad platform delivery metrics. Impressions, clicks, reach and spend are logged by the ad platform. Consent affects what happens after the click, not the click itself.

Here is the practical move. Rebalance the report so the top of it leans on consent-immune sources and the conversion section carries an explicit note about measurement coverage. You are not hiding the gap, you are putting the client's eyes on the numbers that still mean what they used to mean. That is also why a cross-platform report beats a single GA4 view right now: when one source narrows, the others hold the story up. You can see how we assemble that mix on the how it works page.

What our own client data shows

We pulled these figures from ReportsMate's production database in October 2026. Fair disclosure: this is our own book of business, it is a small base, and it skews towards boutique agencies managing local and small-business clients. Treat it as directional evidence of where consent changes bite, not as an industry benchmark.

Google Analytics is the most-connected reporting source we see. Across 80 active platform connections spanning 43 connected client accounts: Google Analytics 30 (roughly 7 in 10 clients), Meta Ads 16 (around 37%), Google Ads 13 (30%), Search Console 11 (26%) and Google Business Profile 10 (23%). The source most exposed to consent is the one almost every report is anchored to.

Over half of our connected clients run paid media alongside it. 28 of 43 have at least one paid-ads platform connected, and 15 of 43 have Google Analytics plus a paid platform in the same report. That combination is precisely where the discrepancy surfaces, because the client can see platform-reported conversions and GA4-reported conversions side by side on one page. If that is your situation, Google Ads vs GA4 conversions walks through why the two will never agree, consent or no consent.

Most inherited analytics setups measure no outcomes at all. Of 20 cached GA4 and Google Ads report snapshots, 14 returned zero conversions for the period. That sample covers only two agency accounts, so read it as a pattern rather than a statistic. The point it makes is still the important one: modelling cannot estimate conversions for a property that never defined a conversion. Consent mode is the second problem. An unconfigured key event is the first, which is why we start with GA4 key events vs conversions before touching anything consent-related.

Reporting cadence is settling into weekly and monthly. Of 78 enabled report schedules, 33 are weekly, 33 monthly and 12 daily. That matters here because modelled data and consent rates are noisy day to day. A daily report invites a client to react to a wobble; a weekly or monthly one gives the pattern room to be real.

And the report has to be read to do any of this work. Across 3,168 tracked report emails, around 28% were opened, the median open landed about 4.7 hours after sending, roughly a third within the first hour and about three quarters inside 24 hours. A consent explanation written into the body of an email that lands in an inbox gets read in an afternoon. The same explanation parked behind a dashboard login waits for a client who may never log in.

How to explain consent mode data gaps to a client

Lead with the cause, not the number. A client who hears "your traffic dropped" braces for bad news. A client who hears "we changed how we count, here is what the new count means" is being brought into a decision. Same data, completely different meeting.

A sequence that works:

1. Name the date. Tie the step change to the day the banner or consent configuration went live. A dated cause converts a scary chart into a known event.
2. Separate reach from measurement. Show Search Console clicks or ad impressions over the same window, holding steady. This is the single most reassuring slide you can put in front of a client.
3. State which mode is running. "Your site uses advanced consent mode, so Google estimates part of the gap" or "your site uses basic mode, so visitors who decline are not counted at all." Precision buys credibility.
4. Be honest about modelling eligibility. If the client's volume is nowhere near 1,000 daily users, say the modelling is not active for them and the reported figures are consented visitors only.
5. Reset the baseline in writing. Agree that pre-banner and post-banner periods are not comparable, and that year-on-year comparisons crossing that date need a caveat for the next 12 months.
6. Switch the KPI conversation to directional. Trends, channel mix and lead quality still work. Absolute session counts compared across the change do not.

One line we hand to agencies, adapt it freely: "Since your cookie banner went live, our analytics only counts visitors who agree to be counted. Your actual audience did not shrink, the measured slice of it did. Search visibility and ad delivery are unchanged, and we will report trends from the new baseline."

Put that paragraph in the report itself, not in a separate email. A caveat that lives next to the chart it explains gets read by whoever opens the report. That is the whole reason we deliver reports as branded emails rather than a login: the commentary and the chart arrive in the same place, in the client's inbox, under your agency's sender identity (your own domain and logo, so it reads as your work and not a tool's). You can see the plans on the pricing page.

A reporting checklist for consent mode v2

Run this once per client, then revisit when anything changes on their site:

• Confirm which consent mode is live. Google Tag Assistant or your consent platform's debug view will show the consent state the tags receive. Guessing here wastes weeks.
• Check consent mode is on every page. Google requires it across all pages and app screens, with tags loading before the dialog appears. A banner missing from the blog or a landing-page subdomain creates a partial, confusing picture.
• Verify the default is set correctly for the client's regions. A default of denied applied globally, when only EEA and UK traffic requires it, suppresses more data than the law asks for.
• Record the go-live date in your reporting notes. You will need it for every year-on-year comparison for the next 12 months.
• Decide if the client is modelling-eligible. Compare their daily user and event volume against Google's published thresholds and write the answer down. If they are not eligible, stop describing their numbers as if a model is filling the gap.
• Audit the key events first. A property with no configured conversions has nothing to model. Fix measurement before blaming privacy.
• Add consent-immune sources to the report. Search Console and Google Business Profile give you numbers that did not change, which keeps the report credible while GA4 narrows.
• Document the caveat in the report template. Write it once, let it ship with every send. A caveat you have to remember to add is a caveat you will forget.

FAQs

Q: Does consent mode v2 reduce my client's GA4 traffic numbers?

A: Yes, in most cases, and the size depends on consent rates and traffic geography. GA4 needs first-party cookies to count users and sessions, so visitors who decline or ignore the banner are not counted directly. The drop concentrates in European Economic Area and UK traffic, where consent is legally required, so a client with mostly non-European traffic sees far less change than one selling into Germany or France. If the site runs advanced consent mode and clears Google's volume thresholds, modelling refills part of the gap. If it runs basic mode or sits below the thresholds, the gap stays visible in the report. Either way the audience itself did not shrink, only the measured portion of it.

Q: Are modelled conversions real conversions?

A: Modelled conversions represent real conversions that happened but could not be observed directly, estimated from the behaviour of consented visitors with similar characteristics. They are a statistical inference, so treat them as a good estimate rather than a counted fact. Google Ads Help distinguishes an advertiser-specific model under advanced consent mode from a general model under basic, so the detail of the estimate depends on the implementation. The practical rule for client reporting: use modelled figures for trend and optimisation decisions, and avoid quoting them to a decimal place in a document that someone will reconcile against their CRM.

Q: Why does Google Ads report more conversions than GA4 after consent mode?

A: The two systems count different things and apply modelling differently, and consent mode widens a gap that already existed. Google Ads credits a conversion to the click that drove it and reports it in the click's date, while GA4 attributes in its own model and reports on the session date. Consent adds a second divergence, because the two products have separate modelling thresholds and separate data sources. A persistent 10% to 30% difference between the two is normal rather than a tracking fault. Set that expectation with any client who audits one number against the other, before they find the discrepancy themselves.

Q: Is Search Console data affected by cookie consent?

A: No. Search Console reports clicks, impressions, average position and queries from Google's own search logs, counted before the visitor ever reaches the client's site. Cookie consent happens on the website, so it cannot affect data that was recorded on Google's side. This is why Search Console has become the reliable organic number in a client report, and also why it will show more organic clicks than GA4 shows organic sessions. That difference is structural, not an error, and it is one of the most common questions clients ask once they start comparing the two.

Q: How do I know whether a client runs basic or advanced consent mode?

A: Check the consent state your tags receive, using Google Tag Assistant or the debug view in the client's consent management platform. Under advanced mode you will see Google tags firing with consent defaulted to denied before any banner interaction. Under basic mode you will see no Google tag requests at all until the visitor accepts. Your client's consent platform settings will usually name the mode directly too. Confirm it rather than assuming, because the answer changes whether you describe the gap in their report as modelled or simply missing.

Q: Should we recommend advanced consent mode to recover the data?

A: Advanced mode recovers more measurement, but the choice is a legal one for your client, not a reporting preference for your agency. Advanced sends an anonymous cookieless ping to Google before the visitor has made a choice. Google's position is that this falls outside consent requirements because nothing is read from or written to the device, and several European regulators and privacy lawyers treat that as unsettled rather than resolved. Present the measurement trade-off clearly, note that it is a decision for their legal adviser, and document whichever mode they choose so your reporting commentary stays accurate.

Q: What should we change in our client report templates because of this?

A: Add a dated measurement note, promote consent-immune sources, and stop comparing across the banner go-live date without a caveat. In practice that means a short standing paragraph explaining what the analytics figures do and do not cover, Search Console and Google Business Profile given real space rather than an appendix, and trend commentary doing the work that raw session counts used to do. Build it into the template once so it ships automatically with every send. Write the commentary layer once, review it when the client's consent setup changes, and leave it alone the rest of the year.

Q: Does modelled data appear everywhere in GA4?

A: No, and this is the detail that causes the most internal confusion. Google Analytics Help states modelled data feeds standard reports, affecting users, sessions and new users, but is not applied to audiences, user explorer, cohort and user lifetime explorations, segments with a sequence, retention reports, predictive metrics, or data exports including BigQuery. So a figure you pull from BigQuery is observed-only, while the same figure in the GA4 interface may include modelling. If your reporting runs off an export and your client reads the interface, you will disagree with each other forever until someone names the reason.

Final tips

Consent mode v2 did not break your reporting. It ended the era where an agency could hand over a number and let the client assume it was a complete census of their audience. The agencies handling this well are not the ones with cleverer tracking. They are the ones whose reports explain what the numbers cover, in writing, every single time.

Three things worth doing this month:

Date the change for every client. One line in your notes, the day their consent configuration went live, saves you a dozen awkward conversations across the next year of comparisons.

Move the report's centre of gravity. Search Console and Google Business Profile did not change. Give them more room, and let GA4 carry trend and behaviour rather than the headline count.

Ship the explanation with the numbers. The consent mode v2 reporting impact is a communication problem wearing a technical costume. A caveat your client reads is worth more than a dashboard annotation nobody opens, which is exactly why our reports land as branded emails under your agency's own sender identity instead of waiting behind a login.

Stop losing your Sundays to client reports. Start your free 14-day trial - no credit card, no setup, cancel anytime. Connect GA4, Google Ads, Meta, Search Console and Google Business Profile, write your consent caveat once, and let every client get a branded report in their inbox on schedule.

Sources: Google Analytics Help, behavioural modelling and consent mode, Google Ads Help, about consent mode, Search Console Help, performance report data. First-party figures queried from ReportsMate's production database in October 2026 and described above with their sample sizes.