Marketing Attribution Models for Small Teams
Oct 26,2026
By SEO ANALYSER
Marketing analytics guide
Three attribution models remain in GA4. Here is how each assigns credit, which one suits a small team, and why consistent tagging matters more than the model.
Short answer
With GA4's three remaining models, a small team's real choice is which one to report on and how to cross-check it
GA4 retired first click, linear, time decay and position-based attribution in November 2023, leaving data-driven, paid and organic last click, and Google paid channels last click. A workable rule for small teams, offered here as a recommendation rather than Google guidance, is to report on paid and organic last click while volume or channel count is low, and to compare it with data-driven credit once several channels run at once. Whatever the choice, tagging discipline decides how reliable either model is.
Models
What each remaining model does with credit
One channel takes all the credit
Paid and organic last click gives full credit to a single channel and ignores direct visits. Google paid channels last click falls back to paid and organic last click when a path has no Google Ads click.
Credit follows observed paths
It applies machine learning to converting and non-converting paths, and each model is specific to an advertiser and a key event. Credit can be fractional, and conversions can be reattributed for up to seven days afterwards.
Because the rules-based models are gone, first-touch questions now need another source, such as the paths report or a self-reported field on lead forms. Google's stated reason for retiring them was that fixed formulas cannot keep up with changing customer journeys.
Choosing
Which model fits which small-team situation
The table is a recommendation built on Google's documentation of how each model behaves, not an official rule.
| Situation | Model to report on | Reason |
|---|---|---|
| One or two channels, short path to conversion | Paid and organic last click | Little credit to share, and the result is easy to audit |
| Several channels running together, steady conversions | Data-driven, checked against last click | Credit is shared on observed paths, so shifts versus last click show assisting channels |
| Reconciling with Google Ads figures | Last click for paid and organic | Analytics uses last click for Google Ads conversions based on key events |
| Very few conversions per month | Last click plus path review | A learned model is hard to sanity-check on thin data (inference) |
Tagging
Why UTM discipline matters more than the model
Neither model can repair inconsistent tagging. UTM values are case sensitive, so two spellings of one campaign become two rows, and Google advises one source per platform, one medium per channel and lowercase throughout. A second trap is partial tagging: when click identifiers cannot be used as intended and any UTM parameter is present, Analytics takes all cross-channel source dimensions from the UTM parameters alone, so a link carrying only one parameter reports the others as not set. The practical rule is to set source, medium and campaign on every tagged link, ideally with campaign ID and source platform too.
Process
How to settle on a model in five steps
- 01
Name the decision the report informs
A budget split between paid channels, a content investment and the ROI reporting covered in the guide to data-driven SEO each need a different level of precision.
- 02
Audit channel tagging first
Check source, medium and campaign values for duplicates and not-set rows before comparing any model.
- 03
Compare models on one key event
Use the attribution models report with the same key event and date range, and note which channels gain or lose credit between last click and data-driven.
- 04
Read assisting channels in the paths report
The middle segment of each path, covering roughly the central half of touchpoints, shows channels that assist rather than close. For judging the role of social beyond last click, see the guide to social media analytics.
- 05
Fix one reporting model and record the date
A model change rewrites historical reporting, so the chosen model and the date it was set belong in the reporting notes. Review the choice each quarter rather than monthly.
Limits
What attribution cannot settle
Direct visits generally get no credit, so a channel that prompts later direct visits is likely undervalued by every GA4 model. Engaged views count only under specific conditions, and data-driven conversions can be reattributed for up to seven days, so recent figures can still move. Attribution also describes credit, not cause: it does not show what would have happened without a channel, which is why results used in budget decisions need a sanity check against independent evidence.
State the attribution model, lookback window and date set in the report footer, so that a change in credit is not mistaken for a change in performance.
SEO Analyser
Read channels and search data together
SEO Analyser's Traffic Analytics module puts GA4 and Search Console data in one dashboard, with a seven-channel breakdown and engagement metrics compared across channels. Attribution models are still configured in GA4. Free for 7 days.
Start free trialFrequently asked questions
Attribution model FAQs
Which attribution models does GA4 still offer?
Three: data-driven, paid and organic last click, and Google paid channels last click. First click, linear, time decay and position-based were retired in November 2023.
Does data-driven attribution give credit to direct visits?
Generally no. Direct visits receive no credit unless the entire path to the key event consists of direct visits.
Does changing the reporting attribution model change past data?
Yes. A model change applies to historical and future data, whereas a change to the lookback window applies only going forward.
Why do UTM values need a naming convention?
Values are case sensitive, so differently spelled variants of one source or campaign are reported as separate rows.
Can the paths report show which channels assist conversions?
Yes. It shows channels that initiate, assist and close key events, and the middle segment of each path is where assisting channels appear.
Summary
Pick one reporting model, tag consistently and cross-check it against the other model. Paid and organic last click suits simple setups, data-driven suits several concurrent channels, and the paths report fills the gap left by the retired rules-based models.
References
Sources and further reading
- GoogleGet started with attribution (Analytics Help)
- GoogleAttribution models report (Analytics Help)
- GoogleAttribution paths report (Analytics Help)
- GoogleSelect attribution settings (Analytics Help)
- GoogleAbout the deprecated attribution models (Google Ads Help)
- GoogleURL builders: collect campaign data with custom URLs (Analytics Help)
- GoogleTraffic-source dimensions, manual tagging and auto-tagging (Analytics Help)
- SEO AnalyserTraffic Analytics
- SEO AnalyserData-Driven SEO Guide
- SEO AnalyserData-Driven Social Media Analytics
