Attribution models: six ways to split the credit, and who each one flatters
An attribution model is the rule that decides how much credit each marketing touch gets for a sale.
The rule is a choice, and every choice flatters some channels and starves others.
Why it matters
The model decides what your reports say each channel earned, which decides where the budget goes. Change the model and the same month of sales produces a different ranking of channels, with no change in what any customer actually did.
How it works
Every model starts from the path of touches a customer took before buying. Here is one path to use throughout, with made-up numbers. Four touches over two weeks, then an order worth £100.00.
- A TikTok ad (the first contact)
- An organic search click
- An email click
- A paid brand search click (the last touch before the order)
Here is how each model splits the £100.00.
| Model | TikTok ad | Organic search | Brand search ad | |
|---|---|---|---|---|
| Last click | 0.00 | 0.00 | 0.00 | 100.00 |
| First click | 100.00 | 0.00 | 0.00 | 0.00 |
| Linear | 25.00 | 25.00 | 25.00 | 25.00 |
| Time decay (illustrative weights) | 6.67 | 13.33 | 26.67 | 53.33 |
| Position based | 40.00 | 10.00 | 10.00 | 40.00 |
| Data driven | set by the model | set by the model | set by the model | set by the model |
Last click gives all the credit to the final touch. It needs no judgement, because the last touch is the easiest one to observe. It flatters brand search and retargeting, which sit at the end of nearly every path, and it starves whatever started the path. There is a fuller treatment on the last click attribution page.
First click gives everything to the first touch. It flatters prospecting channels and starves everything that closed the sale. It is useful as a second view beside last click, because the two together show both ends of the path.
Linear splits the credit evenly, so four touches get £25.00 each. It flatters nobody in particular. A glance at a banner counts the same as the click that completed the order.
Time decay gives more credit the closer a touch is to the sale. The usual form halves a touch's weight for every fixed interval further back in time. In the table the weights are 1, 2, 4 and 8 (each touch worth twice the one before it), which total 15, so the TikTok ad gets 1/15 of £100.00, which is £6.67, and brand search gets 8/15, which is £53.33. It flatters the closing channels less brutally than last click does, and still leaves the opener with the smallest share.
Position based gives a fixed share to the first and last touches and splits the rest across the middle. The split used in the table is 40% to each end and 20% across the middle, so the TikTok ad and brand search get £40.00 each and the two middle touches £10.00 each. It flatters the opener and the closer at the expense of whatever happened in between, which in this path is email.
Data driven has no fixed rule. The model compares paths that include a given touch with paths that don't and gives credit in proportion to the difference in outcome it finds. Google Analytics 4 and Google Ads both offer a version of it. It needs volume to work and you cannot check its arithmetic, which the data driven attribution page covers.
What it can't tell you
None of these models knows whether a touch caused anything. They are rules for dividing credit among touches that were observed, and a touch can be present on every path without having changed a single decision. Brand search is the clearest case: the customer was looking for you by name. Only an incrementality test can say whether the click did any work.
They also only see touches that leave a trace. A podcast ad, a recommendation from a friend or a visit on a different device leaves nothing in the path, so no model, however clever, can credit it.
When to use it
Use last click when you need a model everyone can understand and numbers that match the platforms. Add first click beside it when you suspect the openers are undervalued. Use position based or time decay when you want a compromise you can explain in one sentence. Use data driven when you have the volume and are prepared to trust a split you can't audit.
Whatever you pick, pick one and stay with it for a while. A model changed every quarter produces reports nobody can compare.
Tracking it in a spreadsheet
The inputs are conversions by channel under whichever model you use. A scheduled Fresh Sheets report can keep Google Ads conversions current in a tab of its own, each conversion action counted under the model you chose for it. Fresh Sheets replaces that tab's contents on every refresh, so keep the comparison across models on a separate tab.
Related guides
Marketing attribution: what it answers, why it's hard, and what a small team should do
Marketing attribution is the practice of working out which marketing activity led to each sale, so that the channels that earned it get the credit.Last click attribution: the default model, and who it flatters
Last click attribution gives all the credit for a sale to the final marketing touch before the order.Data driven attribution: what it does, what it needs, and why you can't check it
Data driven attribution is an attribution model that splits the credit for a sale across the touches on the path using a rule learned from your own conversion data, rather than a fixed rule like last click or linear.