Consent Mode and Signal Availability

Ingress

Consent Mode often appears to be a functioning implementation, even though part of the signal is missing, part relies on modeling, and the report gives a more comprehensive picture of the situation than reality warrants. The problem is not limited to reporting — it also affects what kind of signal remains available for analytics, advertising, and automation.

Introduction

A company can lose usable data without reporting appearing broken. This is the business problem of Consent Mode. The person responsible for marketing still sees traffic, events, and conversions. The company’s decision-makers receive a report that looks stable enough to serve as a basis for decision-making. Advertising platforms receive data for ad optimization, and no obvious problem is visible.

A report often looks high-quality at the same time as part of the signal essential for decision-making has already been lost.

Because lawful cookie consent restricts data availability, the report no longer describes the full reality with the same precision as in a situation where per-user observability remains broader. Due to consent denials, a great deal of behavioral information falls outside observation, and this information is supplemented through modeling. It is also possible that data cannot be collected in a form that analytics, optimization, and budget allocation genuinely need.

The business significance of Consent Mode is not solely lawful consent management. The question to investigate is how much usable data remains available to the company as a basis for decision-making. If this overall picture is unclear, the company may misinterpret demand, misjudge the impact of marketing, and allow automation to optimize on the basis of too weak visibility.

Consent Mode belongs to the availability layer of measurement architecture. It determines how much of actual behavior remains available to systems. This is demonstrated in the article’s examples. A basic implementation cuts off signal entirely, while an advanced implementation leaves systems with a limited signal.

Consent Mode is not a consent tool but a mechanism that receives the consent signal transmitted by the CMP or banner and modifies tag behavior based on it.

Consent management and a typical interruption in data collection

Example 1: The basic state of cookie consent blocks data collection entirely

Denying consent does not always mean the same as not collecting data. In this example, however, the effect is absolute. No analytics data is collected at all.

This is a functioning measurement where data collection has been intentionally blocked. The tracking codes required for measurement are technically placed correctly on the page, but analytics does not receive data. The situation is resolved by whether the site’s tracking code is given permission to send a signal through to analytics.

In this implementation, when the user denies analytics and advertising consent, the consent remains in a denied state — meaning all tracking is blocked. As a consequence, analytics and advertising tags do not fire. No requests are sent, and no analytics or advertising cookies are set. In practice, only cookies essential for site functionality remain active, such as those related to consent and language preferences.

From a business perspective, this is a more significant matter than a technical implementation detail. When data is not collected, that visit to the web service does not end up available for analytics. In that case, this data cannot serve as a basis for behavioral modeling in Google Analytics 4 (GA4). Nor is the data available to support conversion modeling in Google Ads in the same way as in an implementation where a limited signal can be transmitted even without cookies.

This makes measurement structurally partial. Reporting appears normal in this situation because data accumulates from users who have given consent. At the same time, however, a significant portion of traffic falls outside tracking. What remains is tracking and reporting where only a part of actual demand and behavior is visible.

The core of this example is that cookie consent has been implemented as required by law and is technically correct. From the perspective of tracking availability, however, this means a complete data blackout for users who do not give consent to the use of cookies.

Example 1: What does the situation look like from a tracking perspective?

In this implementation, the analytics tag does not fire and no analytics signal is generated. Measurement is not broken — it is blocked. The report may still appear functional, even though visibility is based only on the portion of traffic that has given consent.

Example 2: Advanced consent management allows limited signal use in tracking

Denying cookies means that cookies may not be used for tracking. It does not, however, mean that no tracking data is collected at all. In this example, the effect on data quality is likely positive. The starting point is that analytics and advertising data collection is blocked, but a limited signal is left available to tracking systems.

This is a deliberately implemented approach where tracking transmits limited information without analytics cookies. The user is not tracked at the same level as a user who has given consent, but this solution preserves visibility into their activity.

From a business perspective, this is a major improvement and an essential difference compared to the situation where data is not collected at all. In the advanced implementation, the consent state remains in a denying denied state for analytics and advertising, but at the same time, limited cookieless pings are sent to Google. In this case, no normal analytics or advertising cookies are created in the browser. The signal sent to Google does not mean full measurement. Limited data is transmitted to GA4 without normal observability based on analytics cookies.

From a business standpoint, this is a different matter than a complete data blackout. When a limited signal remains available, systems have more basis for estimating missing visibility than in a situation where no analytics signal is generated at all. This does not make the data complete. Nor does it turn modeled data into observed data. It only means that visibility into the user’s activity does not disappear in the same way as in the basic implementation.

Denying cookies therefore does not automatically mean complete darkness. The implementation approach of consent management determines how much usable data can be collected for tracking.

Example 2: The absence of consent does not always block all signals

In this implementation, the absence of consent does not cut off the signal entirely. Systems retain limited visibility without normal measurement based on analytics cookies. A request visible on the network does not mean full observed data, but it can serve as a basis for modeling. The limited signal may include, for example, a timestamp, browser information, and referrer information, but it is not tied to a normal analytics cookie or a persistent user identifier.

Why is the problem not detected?

The problem is not detected because everything appears to be working correctly in principle. Data appears normally in reports, cookies are set on the page, and the cookie consent banner is displayed as expected. In this situation, measurement is easily assumed to be in order. That is often an incorrect conclusion. In reality, no report unambiguously tells how large a portion of actual behavior has been observed and how large a portion is missing.

The significance of modeling may also be misunderstood. Modeling can supplement missing visibility, but it does not make data complete or more reliable. Modeled data does not appear in all reports and features in the same way. For this reason, a report may look intact but deviate from reality.

The risk of errors increases when different reports are compared with the assumption that they are based on consistent data. This does not always hold true. One report includes modeled data, while another report is based only on actual events. Yet the reports may be used with equal weight as a basis for decision-making.

The biggest error is not technical but conceptual. Consent Mode is treated too often as a mandatory implementation or a legal requirement, even though it directly determines signal availability. From a business perspective, what is decisive is how much usable signal remains available to systems as a basis for decision-making and optimization.

The case examples show why a “denied” state can lead to two entirely different outcomes: a complete blackout or limited visibility.

What does Consent Mode v2 mean for business?

Consent Mode v2 was implemented so that consent state can be communicated to Google with greater precision — for example, for advertising purposes. From a business perspective, the benefit is more precise consent state management and a better opportunity to retain usable signal. This difference is visible in the second example.

The absence of consent does not block all signal — part of visibility can be preserved. In Google’s own classification, these correspond to basic implementation and advanced implementation. In the former, signal is cut off without consent. In the latter, a limited cookieless signal remains available to systems as a basis for modeling.

What is the business benefit of modeled data?

Modeled data is a statistical estimate generated by Google of behavior or conversions that the system cannot directly observe. In a Consent Mode environment, this relates to situations where the user does not grant analytics or advertising consent. In Google Analytics 4, this appears as behavioral modeling.

In practice, modeled data is the best estimate of missing visibility — that is, missing data. It can affect, for example, user counts, user paths, and conversions. Google describes in its documentation that the purpose of modeled conversions is to estimate conversions that cannot be directly observed. Behavioral modeling, in turn, estimates the actions of users who do not accept analytics cookies. The modeling is based on observed data and the limited signal that can be transmitted even without analytics cookies. Google does not publish the exact methodology of the modeling.

Consent Mode enables modeled data because, through its mechanism, limited information about user activity and consent state can be sent to Google even without analytics cookies. This information is used as the basis for modeling missing visibility.

However, modeled data is not the same as observed data. Nor is it a guarantee of complete data. What improves its quality is Google’s approach: Google adds modeled conversions to reporting only when model confidence is sufficiently high. Modeling is not presented if there is insufficient data to produce a reliable estimate.

The advanced tracking example illustrates this in practice. In the situation where a limited signal remains available, missing visibility can be estimated more reliably than in the situation where signal is cut off entirely.

How do I check whether a GA4 report contains modeled data?

It is possible to check from a Google Analytics 4 (GA4) report whether it contains modeled data. In the GA4 reporting view, an icon appears after the report name that indicates whether the report includes modeled user data. If the report contains modeled data, a more detailed explanation of the nature of the modeled data becomes visible beneath the icon.

In GA4, modeled data does not appear as a separate reporting layer — instead, it is integrated into reports alongside observed data. This is why the icon indicates the presence of modeling but does not break the report into two separate datasets.

The indicator does not mean full observed measurement, nor does it mean that all figures in the report are modeled. The message communicates that the report may also include estimated data.

Modeling requirements — when is enough data collected?

In a web environment, the prerequisites include that Consent Mode is implemented as an advanced implementation on all pages of the site so that Google tags load before consent, regardless of the user’s consent state. The Google Analytics 4 (GA4) property must collect at least 1,000 events per day (in analytics_storage=’denied’ state) for at least 7 days. Additionally, the property must have at least 1,000 daily users sending events (in analytics_storage=’granted’ state) on at least 7 days within the preceding 28 days.

Meeting the above conditions does not yet guarantee that modeling will activate. Google also applies model quality criteria. If the model quality criteria or prerequisites are not met, modeled data is not displayed in reports.

Example: An online store where signal is insufficient for modeling

Assume, for example, a specialty products online store that receives 400 sessions per day. The site’s CMP solution (Consent Management Platform) is implemented correctly, and approximately 35% of users accept analytics cookies. This means that approximately 140 events per day are sent in a consent state. This volume is not sufficient to meet the minimum threshold of 1,000 events.

As a consequence, Google Analytics 4 (GA4) behavioral modeling does not activate. Reports show 35% of traffic, which comes from users who have given consent. In this case, 65% of users are entirely missing from the data, and GA4 does not supplement the gap with modeling. As a result, the optimization signal is based on incomplete data, and algorithms optimize only according to the behavior of users who have given consent.

How does data availability become a decision risk?

Weaker observability lowers the apparent level of actual demand. Broken or shortened conversion paths can narrow the visible picture of marketing’s impact. The combination of observed and modeled data may still appear intact in a report, even though the signal available for optimization is weaker than assumed. In this situation, reporting, decision-making, and automation are no longer operating in the same reality. This becomes visible quickly in practice. Demand is interpreted as lower than it is, and the impact of channels is read incorrectly. Budget is allocated on the basis of limited information, and advertising platforms’ optimization learns from a weaker signal than the organization believes.

Advertising platforms, automated bidding strategies, and targeting systems need continuous feedback. The stronger and more usable the signal that remains available to them, the better they can learn. Competitive advantage arises from who delivers more permitted and usable data as a basis for systems’ learning.

A technical limitation becomes a risk when incomplete data begins to be used in business decision-making. The basic measurement example shows what a complete data blackout looks like. The advanced measurement example shows how, when cookies are declined, a limited but business-significant signal remains available to systems.

Business impact

Consent Mode affects the quality with which a company is able to interpret demand, manage budget, and steer advertising platform behavior. Therefore, the difference in the article’s practical examples is not a technical detail but a business-level difference in signal quality.

When tracking data deteriorates, the company loses not only data but also part of the information that describes actual markets. In that case, demand appears smaller than it actually is, because part of traffic falls outside reporting. The impact of channels may appear narrower than reality if attribution breaks. Marketing return may appear weaker than reality if part of conversions goes undetected. At the same time, advertising platforms learn to target advertising based on a more limited signal.

At this point, the difference between a fully blocking and an advanced Consent Mode implementation becomes business-significant. Even in the advanced implementation, complete visibility into data is not achieved, but systems retain more usable signal as a basis for analytics, modeling, and optimization. The difference can be reflected in advertising platform learning, audience usability, and budget allocation. In summary, the question is how well the company succeeds in competing for the same users against other actors.

The impact of Consent Mode is indirect but significant. A weaker or incomplete signal weakens the quality of optimization. A higher-quality and legally permitted signal can improve the prerequisites for budget steering. The question is how well the company is able to create competitive advantage through usable optimization signal.

sGTM improves the quality of permitted signal

Consent Mode determines how much signal may be collected. However, it alone does not determine how much of that permitted signal ends up available to systems. In client-side measurement, even data from users who have given consent can be lost due to browser-based restrictions. Browser cookie restrictions such as Intelligent Tracking Prevention (ITP) and ad blockers can shorten cookie lifespans or block tracking code execution, even when the user has given consent.

Server-side Google Tag Manager (sGTM) changes this setup. When tracking operates in the company’s own domain in a first-party context, many browser-based restrictions no longer affect signal collection in the same way. This means that data from consenting users is preserved more comprehensively, and conversion paths do not break for technical reasons.

sGTM does not bypass Consent Mode. If no signal is generated due to the absence of consent, the server side does not create it from nothing. But when signal is permitted, sGTM provides three business benefits that client-side measurement cannot offer:

• Signal loss decreases. Data from consenting users does not disappear due to browser restrictions, which increases the data volume available to advertising platforms and improves the learning conditions for algorithms.
• Data can be enriched with business information. On the server, the value of a conversion can be combined with, for example, actual sales margin or customer lifetime value from back-end systems. In this case, advertising platform optimization can target profitability instead of mere volume.
• The company gains more precise control over what data is transmitted to third parties. The server acts as a filter where data can be anonymized or restricted before being sent onward. This reduces data privacy risk compared to a situation where third-party scripts operate directly in the browser.

Consent Mode is the availability layer that determines signal availability. sGTM is the collection layer that determines how much of that permitted signal is captured and at what quality it reaches systems.

Conclusions

Consent Mode should not be evaluated by asking whether the banner is visible or whether data is flowing to Google. Those do not answer the question that matters for business. What matters is how much usable signal remains available to the company for analytics, modeling, and optimization.

If this whole is treated with indifference, the organization easily makes a classic mistake. In that case, it is imagined that consent acceptance and behavioral measurement are solved with a single technical action. In reality, only partial data collection and utilization is accepted.

Consent Mode is the availability layer of measurement architecture. It determines how much of reality remains available to systems. Partial visibility is not the same as intact measurement, even if the report appears stable. The amount of usable signal directly affects how well the company is able to compete for the same users against its competitors. Consent Mode does not only determine consent management. It determines how much usable signal remains available for analytics, modeling, and optimization.

Further reading and official sources

About consent mode
About consent mode modeling
Consent mode overview
How Google uses consent mode data
[GA4] Behavioral modeling for consent mode

Accessed March 19, 2026