Consent signals and modelling: measuring conversions after cookie refusal (2026)

Imagine a third of your website visitors refusing cookies. Without the right measurement infrastructure, a significant share of your conversion data simply disappears. Your Smart Bidding strategy operates on an incomplete picture, your tROAS or tCPA targets are miscalibrated, and you may be underbidding on traffic that actually converts. This is the daily reality for advertisers in 2026. Consent signals and conversion modelling are the technological solution that structurally addresses this problem, without violating privacy legislation. This article explains exactly how it works, when it matters most, and how AdBrains own AI technology takes this process to the next level.

Why cookie refusal is a measurement problem for Google Ads

Within the European Union, placing non-functional cookies without explicit consent is prohibited under the ePrivacy Directive and GDPR. A substantial share of your visitors ignore or actively refuse the cookie banner. Depending on the industry and banner design, refusal rates typically range between 25 and 45 percent.

For a webshop like ToetsJeKennis.nl or a lead generation advertiser like Clima-Active.nl, this has direct consequences. If 35 percent of traffic refuses cookies, every purchase or quote request from that group is completely invisible to Google Ads. Your conversion column is structurally understated, and your Smart Bidding algorithm believes certain campaigns perform worse than they actually do. The result is lower bids, fewer impressions and lost revenue.

The core of the problem is that classic pixel-based conversion tracking depends entirely on the visitor's consent status. No cookie means no measurement. For years this was the unavoidable trade-off between privacy and data quality. With Googles introduction of Consent Mode v2, that trade-off has largely been resolved, provided the implementation is done correctly.

What are consent signals exactly?

Consent signals are structured, pseudonymised data points that Google receives when a visitor interacts with your website, regardless of whether that visitor has accepted or refused cookies. They contain no personally identifiable information. Instead, they give Google aggregated behavioural signals about the visitor, such as pages visited, session duration and whether a conversion event occurred.

Consent Mode v2 distinguishes four specific parameters that you configure in your Google Tag or via Google Tag Manager:

• ad_storage: Controls whether advertising-related cookies (such as the Google Ads conversion cookie) may be stored. When refused, no tracking cookies are set.
• ad_user_data: Controls whether user data may be sent to Google for advertising purposes, including Enhanced Conversions.
• analytics_storage: Controls whether analytical cookies (Google Analytics 4) may be set for session measurement.
• ad_personalization: Controls whether the visitor may be added to remarketing audiences.

When ad_storage is refused, classic cookie tracking stops. But with Consent Mode v2 active, the gtag or GTM tag still sends anonymous ping signals to Google, so-called cookieless pings. These pings contain no personal data, but they give Google enough information to make a statistical estimate of the actual number of conversions via modelling.

How conversion modelling works technically

Conversion modelling is Googles statistical technique to fill the gap left by cookie refusal. Based on cookieless pings and the behavioural patterns of visitors who did give consent, Google builds a statistical model. That model estimates per campaign, per ad group and per keyword how many conversions likely occurred among the non-consenting group.

The model is based on machine learning and accounts for factors including:

• Historical conversion behaviour per channel, device and time of day
• The ratio between consenting and non-consenting visitors per segment
• Behavioural signals from cookieless pings (page depth, session duration, scroll events)
• Industry-specific conversion rate benchmarks
• Seasonal patterns and external market factors

Modelled conversions are added to your Google Ads reporting, fully transparently alongside measured conversions. Smart Bidding uses both signal sets combined to calculate bids. For E-4motion.com, a webshop for new electric folding bikes offering both direct purchases and test ride requests, this ensures that both e-commerce conversions and leads are fully included in bid optimisation, including those from visitors who closed the cookie banner.

Modelled conversions are not fabrications. They are statistical estimates with an average accuracy of 98 percent relative to actual behaviour, provided the model has sufficient training data. That requires a minimum threshold of measured conversions per campaign, typically at least 50 per month, for reliable modelling. This is precisely why solid first-party data collection is so critical alongside Consent Mode v2.

Consent Mode v2 vs. standard tracking: a direct comparison

To make the added value concrete, consider a lead generation campaign for Clima-Active.nl, an installer of air conditioning and heat pumps. With 1,000 clicks per month and a refusal rate of 35 percent, 350 clicks come from visitors who accept no cookies.

Advertisers switching to Consent Mode v2 with conversion modelling see on average 23 percent more conversions tracked, simply because the measurement gap is closed. That extra data feeds Smart Bidding and leads to better campaign performance without increasing the advertising budget.

The combination with Enhanced Conversions amplifies this effect further. Enhanced Conversions sends hashed first-party data (such as an email address after a purchase or lead form) to Google, allowing the model to match more accurately and further enriching the signal set. Server-side tracking via a dedicated sGTM infrastructure adds another layer: conversion signals are enriched with first-party data before reaching Google, preventing browser-based ad blockers or ITP restrictions from disrupting measurement.

How AdBrains AI automates consent signals and modelling

At AdBrains, consent-proof measurement is not a manual checklist but a fully automated component of our AI infrastructure. While other agencies configure Consent Mode v2 once and never revisit it, our AI monitors the quality and completeness of the consent implementation daily for every account we manage.

Our server-side signal enrichment works as follows: we operate a dedicated sGTM (server-side Google Tag Manager) infrastructure per client. Conversion signals are enriched with first-party data at server level, hashed and then forwarded to Google Ads and GA4. This completely bypasses browser-based tracking restrictions and ensures that every conversion, including those from Safari users with ITP or visitors using ad blockers, still arrives as a signal at Google Smart Bidding.

Our multi-agent verification system additionally checks daily whether the consent signals arriving from a website are consistent with the expected behavioural pattern. If an anomaly is detected, for example a sudden drop in cookieless pings or a contradictory consent status in the data layer, our account managers are automatically alerted before Smart Bidding performance is affected. This prevents a broken cookie banner or an incorrect GTM configuration from going unnoticed for weeks and silently undermining campaign results.

Furthermore, our automatic tCPA/tROAS optimisation system adjusts bidding strategies as soon as modelled conversion data reaches a reliable threshold. For ToetsJeKennis.nl, this means the target ROAS is recalibrated on the combined volume of measured and modelled conversions, not just the measured subset. As a result, bids are never set too conservatively on segments that actually perform well but appear underperforming in raw data due to cookie refusal.

Finally, our AI integrates consent signal status into the Keyword Incubator and auto campaign expansion modules. New keywords tested in the incubator are only promoted to production campaigns when the conversion data, including the modelling component, is sufficiently reliable. This ensures that decisions about scaling or pausing keywords are always based on the most complete view of reality, never on flawed measurement.

The combination of server-side signal enrichment, multi-agent monitoring and automated Smart Bidding calibration means AdBrains clients structurally extract more return from their Google Ads budget, even in a world where more and more visitors refuse the cookie banner.

Practical implementation steps for Consent Mode v2

For advertisers who want to implement consent signals and modelling themselves, here are the essential steps for a correct setup:

1. Install a Consent Management Platform (CMP) certified by Google as a CMP Partner, such as Cookiebot, Usercentrics or OneTrust. This ensures a technically correct link between the cookie banner and the consent parameters in GTM.
2. Activate Consent Mode v2 via Google Tag Manager by configuring the official Consent Initialization trigger and setting all four parameters to the default status "denied" at page load, before consent is given.
3. Configure the update trigger so that consent parameters are automatically updated to "granted" when a visitor accepts cookies in the CMP.
4. Implement Enhanced Conversions in Google Ads by sending hashed user data (SHA-256 email addresses) with conversion events. This enriches the model and improves the accuracy of modelled conversions.
5. Validate the implementation using the Consent Mode debugger in GTM and check in Google Ads whether modelled conversions appear in the conversion report.
6. Consider server-side tracking via sGTM for maximum signal integrity, particularly if your audience includes many Safari or Firefox users, or if you operate in a sector with high cookie refusal rates.

Frequently asked questions about consent signals and conversion modelling

What is the difference between Consent Mode v1 and Consent Mode v2?

Consent Mode v1 introduced the basic parameters ad_storage and analytics_storage, allowing Google to adapt cookies based on consent status. Consent Mode v2, mandated by Google from March 2024 for use in the EEA, adds two new parameters: ad_user_data and ad_personalization. These are specifically required for sending user data to Google for advertising purposes and for building personalised ad audiences. Without the v2 parameters, you lose access to certain Smart Bidding features and audience segments, even if the visitor has given consent.

Are modelled conversions as reliable as measured conversions?

Modelled conversions are statistical estimates, not direct measurements. At the aggregated level, across campaigns and time periods, estimates average 98 percent accuracy relative to actual conversion behaviour. At the individual level, they cannot be traced back to a specific person or session. For Smart Bidding optimisation, they are fully equivalent to measured conversions; the algorithm treats both signal types equally. Reliability increases as a campaign generates more conversion volume and the first-party data foundation becomes stronger.

How do I see modelled conversions in my Google Ads account?

In Google Ads, you can add a specific column for "Modelled conversions" or "Including modelled conversions" to your conversion columns. In the default view, the "Conversions" column already shows a combination of measured and modelled conversions, provided Consent Mode v2 is correctly implemented. In GA4, the reporting allows you to see segmentation between measured and modelled sessions. Via the Google Ads API, modelled conversions are also available as a separate metric, which is useful for advanced dashboards and reporting tools.

Is conversion modelling only relevant for the European market?

Conversion modelling is technically available globally, but its relevance is greatest in markets with strict privacy legislation such as the EEA (GDPR/ePrivacy), the UK (UK GDPR) and certain US states (CCPA in California). In markets without mandatory cookie banners the refusal rate is lower, but iOS-related tracking restrictions (ITP in Safari) create structural measurement gaps outside Europe as well. Conversion modelling is therefore a universally relevant technique for any advertiser relying on Google Ads for performance measurement.