Last-click attribution
Last-click attribution is the rule that gives 100% of the credit for a sale to the final click a customer made before buying. Every earlier ad, email or visit in the journey gets nothing. It is the oldest and simplest attribution model, and still the default view in many ecommerce reports, so the channels that close sales look strong and the channels that start them look weak.
| Variable | What it covers |
|---|---|
| Final click | The last tracked click before the order. Views, impressions and earlier clicks are ignored. |
| Order value | The revenue of the order, credited in full to that one channel. |
| Lookback window | How far back a click can sit and still count, set per tool. Change it and the credit moves. |
| Direct visits | Some tools skip a final direct visit and credit the click before it, a variant called last non-direct click. |
Worked example
Worked example / demonstrative numbers. Branded search tops the table because it sits at the end of journeys that other channels started.
Is last-click attribution good enough?
It is good at one thing: it is simple, consistent and hard to argue with mechanically. Everyone can see which click came last, and the totals add up to the tracked revenue, with no weighting opinions involved. For a store with short, single-visit journeys and one or two channels, that can be enough.
As soon as buyers see several channels before purchasing, last click systematically over-rewards demand capture (branded search, email, retargeting) and under-rewards demand creation (prospecting video, social, creators). Budget follows credit, so a business run on last click tends to cut the channels that fill the top of the funnel, then wonders months later why branded search volume fell. It also says nothing about cause: the final click on a branded search ad may have been a customer who would have typed the URL anyway. Compare it against multi-touch attribution, and test the big calls with incrementality. The wider argument is covered in who gets the credit.
Last-click attribution vs related models
| Model or metric | How credit is given | How it differs from last click |
|---|---|---|
| First click | 100% to the first tracked touch. | The mirror image: over-rewards discovery, ignores the close. |
| Multi-touch attribution | Split across all tracked touches by a rule. | Assisting channels get some credit instead of none. |
| Data driven | Weights fitted by the platform's algorithm. | Less arbitrary than last click, still correlational and still limited to tracked touches. |
| Holdout test | No credit rule: compares exposed and unexposed groups. | Measures what the channel caused, which no attribution rule can. |
| MER | Total revenue ÷ total spend. | No attribution needed, so last-click bias cannot creep in. |
Common mistakes
- Cutting prospecting because its last-click ROAS is low. Prospecting rarely holds the final click. Judge it on new-customer CAC and a holdout, not on closer credit.
- Scaling branded search on its last-click ROAS. Much of that demand already existed. Its ROAS looks superb partly because it sits last, per what a good ROAS is.
- Comparing tools with different windows. A 7-day and a 30-day last-click report credit different orders. Line the windows up before comparing channels.
- Mixing last click with platform self-reporting. Your analytics tool uses last click; each ad platform uses its own model and window. Their totals will never agree, and summing them double counts, as platform over-claiming shows.
- Reading revenue credit as profit. Last-click revenue ignores discounts, shipping and COGS. Put the credited orders through CM1 to CM3 before judging a channel.
Last-click attribution FAQ
Related
Inside Blufire, S5 Acquisition reconciles attribution models in one comparison matrix, so last click is one visible view beside the others rather than a choice silently baked into every dashboard number.
Updated September 2026