Marketing attribution for games and websites
See which channels, campaigns and creators lead to signups, wishlists and purchases. Compare six rule-based models, and follow one player from the website into the game.
How to install ↗Six rule-based models
Pick a conversion event, then choose how credit is split between the touches before it. Compare two models side by side to see how the credit moves.
| Model | How credit is split |
|---|---|
| Last non-direct (default) | All credit to the last touch that is not Direct. When all touches are Direct, the last touch. |
| First touch | All credit to the first touch in the lookback window. |
| Last touch | All credit to the last touch before the conversion. |
| Linear | Equal credit to every touch in the lookback window. |
| U-shaped (40/20/40) | 40% to the first touch, 40% to the last touch, 20% split between the touches in between. |
| Time decay | Touches nearer to the conversion get more credit. Credit halves for each 7 days before the conversion. |
What counts as a touch
A touch is the first page of a visit, a page whose source changes during a visit, or a referral code that a player types. The source comes from UTM parameters, ad click IDs, referral codes and the external referrer that the SDK records.
Choose a lookback window of 30, 60, 90, 180 or 365 days. The report starts at 90 days, because a game sale often comes months after the first visit. A conversion with no touch in the window is in the Unattributed row; it is not counted as Direct.
Group the credit your way
Show credit by the dimension that matches your question.
- Channel (for example Organic Search, Paid Social, Email, Referrals)
- Source, medium and campaign from UTM parameters
- Creator, from referral codes
- Short link, from tracked links
- Ad network
- Referrer
One player across the website and the game
A player often finds the game on the website and buys in the game. Attribution follows the player, so the website touch gets the credit for the purchase in the game.
The website and the game join in two ways: both call identify() with the same user ID, or the game opens your website with the Godot SDK's link_url() or the Unity SDK's LinkUrl(), which needs no sign-in. A link from the game works for 10 minutes. A link opened in more than 3 browsers counts as shared and joins nothing. Without either, the browser and the game install count as two players.
Revenue, not only conversions
Each model also splits revenue. Revenue comes from the revenue property on the conversion event, or from Stripe and Paddle payments when the conversion event is purchase. A report uses one currency at a time and does not convert between currencies. Refunds are not subtracted, so the report shows gross revenue.
Export and automate
Download any report as CSV. The same report is available in the Analytics API (scope attribution:read) and as the get_attribution tool in the MCP server, so an AI assistant can compare models for you.
Frequently asked questions
Is there a data-driven or machine learning model?
No. The six models are rule-based, so you can see exactly why a touch gets its credit.
Does attribution include ad spend or ROAS?
No. Manalens does not import ad spend, impressions or view-through data. It credits observed visits and touches.
Can it attribute Steam sales?
Steam does not share which player bought, so Steam sales are not in this report. Import the Steamworks UTM report to see Steam visits, wishlists and purchases by campaign on the Steam report.
Does it work for players who never sign in?
Yes, inside one browser or one game install. To join the website and the game without sign-in, open the website from the game with the Godot SDK's link_url() or the Unity SDK's LinkUrl().
Start with a clear measurement model.
Install the SDK, then explore traffic and product activity with the same definitions.