Google Analytics session attribution confusion spreads among professionals

A LinkedIn poll posted by Witold Wrodarczyk on December 9, 2024 revealed widespread confusion about Google Analytics 4 session attribution mechanics among advertising and analytics professionals. The poll attracted 624 votes, with 59% providing incorrect answers to a fundamental question about how the platform attributes direct traffic to previous sessions.

According to the poll, Wrodarczyk asked participants to predict session source attribution when a user visits a website from Instagram, then returns two days later by typing the URL directly into the same browser. The majority of respondents selected "(direct)" as the session source, while the correct answer was "instagram.com" based on Google Analytics 4's last non-direct click attribution model for session sources.

The confusion stems from how Google Analytics handles direct traffic differently than many professionals expect. Session-scope attribution in GA4 determines the source that started each session, but the platform avoids assigning sessions to direct traffic whenever possible by inheriting the previous source within the lookback window. The default lookback window in GA4 is 90 days, according to documentation from Search Engine Land published May 26, 2024.

Simo Ahava, co-founder at Simmer and partner at 8-bit-sheep, confirmed the attribution behavior through testing documented in the LinkedIn discussion. On December 8, Ahava collected a single session with source "t.me/WebAnalyst" in GA4's Traffic Acquisition report. After examining the session in BigQuery, Ahava discovered the session had no campaign setting parameters at all—it was a direct session. However, the session was attributed to "t.me/WebAnalyst" source because that represented the most recent non-direct source from the user, visible in the BigQuery table showing the user visited from this source in mid-November.

The discussion revealed a significant problem: Google's own documentation contains incorrect information about session attribution. According to Ben Luong, senior digital analytics consultant, Google Analytics help article 14731736 contradicts the actual platform behavior. The documentation suggests that following sessions after an initial source attribution will be direct, which testing has proven incorrect.

Wrodarczyk conducted verification testing to settle the debate. According to his December 10 comment, he visited his website once with UTM parameters "/?utm_source=test1&utm_medium=test1" and then closed the window. He subsequently visited the website directly by typing the plain URL multiple times with intervals longer than 30 minutes, on the same day and the following day. The results showed four sessions with the test1 source, despite the link with UTM being used only once. This confirmed that session attribution operates in last non-direct mode.

The attribution methodology reflects a fundamental shift from Universal Analytics. In Universal Analytics, default reports attributed entire credit for conversions to the last click using what was called the last non-direct click model. GA4 offers wider availability of different attribution models, but the behavior depends on report scope—whether user acquisition source, session source, or event source.

Session source will only show as direct when Analytics cannot identify any other source of visit for a given user within the lookback window. This creates scenarios where users see sessions attributed to sources from weeks prior, as long as those interactions fall within the 90-day default window. The lookback window determines how far back in time a touchpoint is eligible for attribution credit, with options to change it to 60 or 30 days in property settings.

Google Analytics expanded its core scoping features from one to three dimensions on December 30, 2024, introducing user scope, session scope, and event scope to replace its previous session-based tracking system. This technical modification affects how marketing attribution data is collected and analyzed across different interaction points.

The first user source dimension represents another layer of attribution complexity. This dimension shows where users came from to the website or application for the first time, recorded by the first_visit event for websites or first_open event for apps. According to the Search Engine Land documentation, the first visit source is attributed using the last non-direct click model, applying only to interactions before the first website visit or app open.

Once assigned, the first user source remains unchanged as long as Google Analytics can technically link the user's activity on the website and app with the same user. The first user source will reset if tracking of the user is lost, such as when the user does not visit the website for a period longer than the Analytics cookie expiration date.

The cookie expiration time in GA4 is set to 24 months by default but can be changed to periods between one hour and 25 months. Subsequent visits may renew this time limit. However, the data retention period is set by default to only two months, though this can be changed to 14 months. In the paid version, Google Analytics 360, retention can extend to 50 months.

Event scope attribution introduces additional complexity through attribution models set in GA4 property settings. The default data-driven model can be changed at any time, with changes working retroactively to affect historical data. The platform offers several predefined models including last click, first click, linear, time decay, position-based, and data-driven attribution.

All models use the non-direct principle, meaning they do not assign value to direct visits unless there is no other interaction on the path. The ads-preferred models assign the entire value of key events to Google Ads interactions if they occur in the funnel. The controversial Google Paid Channels Last Click reporting attribution model attributes 100% of conversion value to the last Google Ads channel a customer interacted with before converting.

The LinkedIn discussion highlighted how professionals with extensive industry experience struggled with these mechanics. Krista Seiden, founder of KS Digital and former Google employee, commented on December 10 that the results were "incredible how skewed (in the wrong direction)" they were. She noted shock at the job titles of those who provided wrong answers.

Grant Higginson, digital analytics and marketing engineering consultant, initially argued both answers could be correct depending on interpretation. According to his December 10 comment, the source of the second session for the user under user acquisition reports would be Instagram, but for traffic acquisition reports it would depend on whether the question referred to the specific session or the user's overall source attribution.

Wrodarczyk addressed this interpretation in his response, clarifying that first user source is assigned to the user "forever" in Analytics reports. This means it is possible for users to have their first visit from another source more than 90 days ago and then have a direct visit. In such cases, these visits will show as direct because the visit from the other source reflected in the first user source was beyond the lookback window.

Google Analytics boosts Data Import features and fixes Attribution Bug on September 25, 2024, addressing an issue affecting session attribution when using user-provided data. The problem caused over-attribution of sessions to direct traffic under scenarios involving user-provided data with automatic user data collection.

The discussion also revealed limitations in how Google Analytics handles sessions during active browsing. In GA4, a session begins when a user visits the website or app and ends after user inactivity for a specified time—30 minutes by default. Closing the browser window does not end the session. If the browser window is closed, another visit to the website within the time limit will still belong to the same session unless the browser deletes cookies and browser data after closing.

If a visit from a new source occurs during a session, a new session will not start, and the source of the current session will remain unchanged. This does not mean the visit from the new source is ignored. GA4 records the source of this visit, and event-scope attribution reports will take into account all sources of all sessions.

Rethinking conversion windows for accurate attribution in GA4 examines how the default settings for conversion windows might not be ideal for achieving accurate attribution data. Within GA4, a conversion window defines a time frame measured in days that determines how far back in time a touchpoint can be credited for influencing a conversion event.

The attribution paths report in GA4 provides rich data including days to key event and the number of interactions for a given path, referred to as touchpoints to key event. This partly compensates for the lack of time lag and path length reports that were separate in Universal Analytics. The ability to choose an attribution model for this report may be surprising at first sight, as the attribution model does not affect attribution paths themselves but rather how credit is distributed across interactions in the selected model.

Modeled and blended data represents another fundamental change in Google Analytics 4. Reports include data collected by tracking code enriched with modeled data using information collected in cookieless consent mode for users who have not given consent to tracking and data for users logged into Google. This data is fragmentary, but Google can fill in the missing data using extrapolations and mathematical modeling.

Google Analytics refocuses user-provided data on ads conversions on November 5, 2025, shifting the system's focus toward advertising conversion accuracy and away from user session attribution. The infrastructure update represents a fundamental change in how the platform processes first-party customer information.

The poll controversy highlighted growing concerns about platform documentation accuracy and the challenges professionals face keeping current with attribution mechanics. Multiple participants noted that even Google's official help articles contradict actual platform behavior, creating confusion for marketers attempting to implement proper tracking and interpret campaign performance.

Jeremy Muratore, a marketing professional focused on crafting marketing at scale, commented in the discussion that the distinction between different source dimensions creates complexity. A source is not universal anymore in GA4, with variations including ad source, CM360 source, DV360 source, SA360 source, first user, manual source, manual source platform, session source, and session manual source. UTM parameters have shifted into different buckets enabling complex and confusing attribution modeling conversations.

The technical implications extend to reporting accuracy and campaign optimization. When marketers misunderstand session attribution, they may incorrectly evaluate channel performance, misallocate budget, or draw false conclusions about customer journey patterns. GA4 adds key event rate metrics to User Acquisition and Traffic Acquisition reports, introducing user key event rate and session key event rate to provide deeper engagement insights.

The discussion also touched on how Google Analytics processes traffic from payment gateways and external sites during active sessions. Sources of these visits are considered unwanted referrals and should be excluded in configuration. Visits from excluded referrals are reported as direct visits. In GA4, these visits are ignored because the session source and session count remain unchanged, with the non-direct attribution modeling assigning no credit to the direct source.

Google Analytics updates Attribution Model for Paid Search in June 2024, addressing issues where conversions attributable to paid search were sometimes mistakenly attributed to organic search, particularly for single-page applications. The update focused on capturing campaign information associated with the first user event on each webpage.

The broader industry context shows attribution mechanics becoming increasingly complex as privacy regulations, platform changes, and industry standards reshape digital advertising infrastructure. Meta restricts attribution windows and data retention in Ads Insights API, eliminating two view-through attribution windows and imposing strict historical data retention limits starting January 12, 2026.

Martin Botťánek, a technical marketer with expertise in Google Tag Manager, noted in the discussion that even ChatGPT initially provided a correct answer but required justification based on source documentation. He also mentioned working on CSS for Google Ads in the EU, where many claim it gives a 20% discount. LLMs initially wrote it's true, even hallucinating that Google said this, but then admitted it's just a claim by some CSS used for marketing purposes.

The poll results suggest widespread need for professional development in analytics attribution mechanics. With 59% of 624 respondents selecting the incorrect answer, including individuals holding senior positions in analytics and advertising, the knowledge gap extends across experience levels. The discussion generated 75 reactions and 50 comments, indicating significant interest in clarifying these mechanics.

Wrodarczyk thanked Julius Fedorovicius for sharing the discussion with his audience, noting the conversation wouldn't have happened without that amplification. The poll was tagged with #wwQUIZen, suggesting this represents part of a series of educational quizzes Wrodarczyk conducts on analytics topics.

GA4 strengthens campaign measurement with Floodlight Linking announced in June 2024, expanding linking capabilities to include subproperties and roll-up properties for enhanced campaign measurement across complex business structures. The integration enables conversion sharing, campaign insights, and improved attribution through various attribution models.

The technical specifications documented in the Search Engine Land article emphasize that lookback window changes do not work retroactively, applying only from the moment of the change. Engaged views of YouTube ads, however, always have three-day lookback windows regardless of property settings. This creates scenarios where different interaction types have different attribution windows simultaneously.

The conversion export to Google Ads has flexible options important to understand. In GA4, conversions can be imported using the attribution model selected in Google Ads conversion settings. Additionally, marketers can choose whether GA4 conversion export attributes conversions only to Google Ads or across all channels. Attributing only to Google Ads makes the conversion export similar to native Google Ads tracking, with conversions attributed solely to Google Ads clicks on the attribution path.

The model comparison tool in GA4's Advertising section allows comparison of different attribution models regardless of property-level attribution settings. Currently available models match those in property settings, with no ability to create custom models. GA4 allows reporting in two attribution time methods: interaction time and key event time, each with distinct advantages and disadvantages for different analysis needs.

Google publishes analytics reporting playbook for marketers in October 2025, addressing widespread confusion about the platform's multiple reporting surfaces and their distinct capabilities. The resource provides detailed guidance on Reports, Explore, Advertising, Data API, and BigQuery export functionalities.

The data retention implications affect how long Google Analytics can remember user data. In addition to cookie expiration, GA4 has a data retention period set by default to only two months, though this can be changed to 14 months. After this time, Google deletes user-level data from Analytics servers. To keep this data, it must be exported to BigQuery. In the unpaid version of GA4, first user source data are deleted after 14 months of inactivity, after which the user will be recorded as a new user.

The discussion participants included professionals from various geographic locations and company types. Wajahat Mubashir, a senior web developer, initially claimed GA4 does not carry over the previous session's source, but Wrodarczyk corrected this by noting that BigQuery's session_traffic_source_last_click also works in last non-direct mode.

The verification methodology Wrodarczyk used included adding "unique_key=test1" parameter to identify URLs of pages visited to ensure direct sessions could be identified if they occurred. This parameter does not impact the test concept—the first visit with UTM could be from any source like Instagram, with following visits being direct URL typing.

Lookback windows don't apply for first user source in the sense of attribution window. First user source remains unchanged for the entire user lifetime. The first visit could be four months or even several years ago, and the user will have the same first user source. Session attribution, however, works only within the lookback window of maximum 90 days, meaning direct session source is possible for users with different first user source.

The practical implications for marketers include the need to understand these attribution differences when analyzing campaign performance. Last-touch returns amid new Google features examines how last-touch attribution represents one of the most straightforward methods available to marketers, serving as the standard attribution option in Google Analytics and most analytics tools.

The confusion documented in this poll reveals how technical documentation errors can propagate misunderstanding across the professional community. When Google's own help articles contradict actual platform behavior, professionals who study documentation carefully may actually develop incorrect mental models of how the system works. This creates a paradoxical situation where diligent research leads to wrong conclusions.

Google Analytics tackles (not set) attribution with latest update on July 26, 2024, addressing a long-standing issue with the prevalence of (not set) attribution values in reports. The update focuses on handling automatically collected events, requiring users to have either a config command or the Google tag installed on a web page for these events to be sent.

The broader measurement landscape continues evolving as platforms balance privacy requirements with attribution needs. Meta tests GA4 integration for cross-platform tracking in ads examines how the new GA4 integration feature provides an alternative measurement pathway that operates independently of Meta's native attribution systems.

Timeline

• December 9, 2024: Witold Wrodarczyk posts LinkedIn poll asking about Google Analytics 4 session attribution behavior
• December 8: Simo Ahava conducts BigQuery test showing last non-direct attribution for session sources
• December 10: Wrodarczyk conducts verification test with UTM parameters confirming non-direct attribution model
• December 10: Ben Luong identifies error in Google Analytics help article 14731736 contradicting actual platform behavior
• December 30, 2024: Google Analytics expands core scoping features from one to three dimensions
• November 5, 2025: Google Analytics refocuses user-provided data on ads conversions
• October 2025: Google publishes analytics reporting playbook for marketers
• September 25, 2024: Google Analytics boosts Data Import features and fixes Attribution Bug
• July 26, 2024: Google Analytics tackles (not set) attribution with latest update
• June 2024: Google Analytics updates Attribution Model for Paid Search
• June 2024: GA4 strengthens campaign measurement with Floodlight Linking

Summary

Who: Marketing and analytics professionals responding to a LinkedIn poll posted by Witold Wrodarczyk, CEO at Adequate and online marketing analytics specialist, with participation from industry experts including Simo Ahava, Krista Seiden, Ben Luong, and Grant Higginson.

What: A poll revealing that 59% of 624 advertising and analytics professionals incorrectly answered a question about Google Analytics 4 session attribution mechanics, specifically how the platform attributes direct traffic to previous sessions within the 90-day lookback window using last non-direct click attribution model.

When: The poll was posted on December 9, 2024, with testing and verification conducted between December 8-10, 2024, in the context of ongoing Google Analytics 4 attribution system changes throughout 2024-2025.

Where: The discussion occurred on LinkedIn, involving Google Analytics 4 session attribution mechanics in the Traffic Acquisition and User Acquisition reports, with technical verification conducted through BigQuery data examination and UTM parameter testing.

Why: The confusion stems from Google's own documentation containing errors about session attribution behavior, the complexity of GA4's three-scope attribution system (user, session, and event), and the platform's non-intuitive handling of direct traffic through last non-direct click attribution within lookback windows that can extend up to 90 days.