• PL
  • EN
  • #google #googleads #googleanalytics #consentmode

    How Consent Mode Really Works

    Aleksandra Görlich Aleksandra Görlich

    What is Consent Mode and Why Should It Be Implemented (Is It Necessary for Everyone?)

    Consent mode is a Google solution that allows the adjustment of tag operations based on user consent status. It enables modeling in Google tools, which means supplementing reports with data from users who have not consented to tracking.

    The introduction of this solution is due to the new regulations of the Digital Marketing Act, which require collecting traffic data only from users who have given their consent.

    All websites using Google Marketing Platform advertising tools are obliged to implement consent mode. It is also recommended for traffic analysis tools such as Google Analytics. Consent mode is one of the solutions that allow for consent management. Consent management is a broader concept because it refers to controlling the operation of all marketing tools (e.g., Meta, Bing, Criteo) depending on user consent, not just Google products. The requirements for consent management may vary in different countries and be subject to local regulations, whereas the requirement to implement consent mode applies to Google products. Fortunately, by implementing consent mode, we can also manage consent management.

    Implementation is usually done using Consent Management Platforms (CMP), where banner settings with options to accept or reject cookies are configured.

    How Consent Mode Works

    The system is based on the selection of which types of cookies the user wants to accept during a visit to the website. Upon the first visit, a banner appears with options to accept, decline, or (most often) customize their consent preferences.

    Consent mode on data.rocks website

    It’s best to place the consent banner in the center of the page so that the user cannot bypass it and must make a decision to accept or reject cookies. Currently, Google Consent Mode focuses on two types of cookies: analytical and marketing/advertising cookies. The former allows for website traffic measurement and data transmission to Google Analytics, while the latter is essential for collecting audience lists for remarketing campaigns and conversion measurement.

    If a site lacks Consent Mode and the user rejects tracking, data in GA and Google Ads is lost. Additionally, the cookie banner may completely block GTM.

    When Consent Mode is implemented, some user data can be supplemented through data modeling for those who rejected cookies. According to Google, up to 65% of lost data can be recovered. We published a case study on data modeling here.

    What is Data Modeling and How It Works

    Modeling works only with the advanced version of Consent Mode. When a user rejects tracking, Google collects anonymized data on these interactions. Google documentation explains:

    Behavioral modeling for consent mode uses machine learning systems to model the behavior of users who reject Analytics cookies based on the behavior of similar users who accept them. Modeled data allows for obtaining useful insights from Analytics reports while respecting user privacy.

    In other words, the data that Google cannot record is supplemented based on the recorded data of the majority of users. For the numbers to make sense, certain threshold conditions must be met:

    • The GA4 service records at least 1,000 events daily with the parameter analytics_storage=’denied’ for at least 7 days.
    • The GA4 service has at least 1,000 users daily sending events with the parameter analytics_storage=’granted’ for at least 7 of the last 28 days.

    Once these conditions are met, modeling should automatically activate, and if not met, it will automatically deactivate. In practice, it works for services with around 30,000 users per month. After correctly implementing consent mode, modeling data is loaded into reports approximately one month later. Therefore, if you are reporting data for the previous month, wait more than a few days before reading it, as changes based on modeling may appear slightly later.

    Effects of Modeling – Case Study

    Some time ago, we had the opportunity to verify how modeling works. We present this using the example of an online clothing store visited by around 40,000 users monthly.

    After two months of implementation and full activation of consent mode on the website, we verified the number of granted and rejected consents in the data sent to BigQuery.

    • About 31% of sessions have a rejected tracking status, which is within the norm.
    • Only 23% of purchases have a rejected tracking status. This is typical – users willing to purchase are more likely to accept tracking.
    GA4 and BigQuery data match in terms of the number of events, sessions, and purchases.
    GA4 and BigQuery data match in terms of the number of events, sessions, and purchases.

    How Data Modeling Works in GA4 with Advanced Consent Mode

    With this data, we checked in GA4 whether the modeling worked. We did this by comparing standard report data in two identity versions for reporting purposes:

    • Mixed Identity – includes modeled data
    • Observed Categories Identity – without modeled data

    Our observations were as follows:

    a) When the user accepts tracking, Google collects interaction data with the site without restrictions. For the examined store, this accounted for about 69% of sessions and 77% of purchases.

    b) When the user rejects tracking, Google collects anonymized data about these user interactions. Then, modeling occurs – Google combines this anonymized data and assigns it to traffic sources or devices based on data collected when tracking is accepted. Modeled data in the selected store accounted for about 31% of sessions and 23% of purchases.

    The data increase is at a similar level for most major traffic sources, approximately +37% to +53% in reported users and sessions, and around +28% to +60% in transactions and revenues.

    Traffic sources

    In the case of smaller traffic sources, the change is often more significant (+5% to +100%), which shows that consent mode most strongly supports traffic sources with less data.

    Data verification has shown that modeling in the described store works efficiently, mitigating the potential lack of data from rejected user consents.

    • Thanks to consent mode, purchase reporting in GA4 is approximately 23% more accurate.
    • Thanks to consent mode, session reporting in GA4 is approximately 31% more accurate.
    • The increase in data in most major traffic sources is about +27% to +37% for reported users, sessions, and about +28% to +60% for transactions and revenue.
    • Smaller traffic sources experience a greater data increase (from +30% to +50%).

    It is worth noting that modeling can “return” different amounts of data to the reports. Depending on the quantity and quality of input data, the modeling may, after about a month, load modeled data into the reports, which will together provide nearly 100% coverage of traffic or much less. It may vary in each case depending on the industry and the volume of recorded traffic.

    Do you need help with
    Consent Mode?

    Write to us