Attribution model playground
Assemble a multi-touch customer journey, set a conversion value, and see five common attribution models divide that value across your channels side by side — the fastest way to understand why two reports of the 'same' campaign disagree.
Credit each model assigns, by channel
| Channel | First click | Last click | Linear | Time decay | Position-based |
|---|---|---|---|---|---|
| Paid search | $1,000 | $0 | $200 | $32 | $400 |
| Paid social | $0 | $0 | $200 | $65 | $67 |
| $0 | $0 | $200 | $129 | $67 | |
| Organic search | $0 | $0 | $200 | $258 | $67 |
| Direct | $0 | $1,000 | $200 | $516 | $400 |
Same journey, same $1,000 — five defensible answers. First-click rewards discovery, last-click rewards the close, and the middle touches only get credit under linear, decay, and position-based. The model you choose is a business decision, not a statistical one.
How it works
Every attribution model answers one question — which touchpoints get credit for a conversion — with a different rule. First-click gives everything to the channel that started the journey; last-click gives everything to the channel that closed it. Linear splits credit evenly; time-decay weights later touches more heavily (here each earlier touch is worth about half the next); position-based gives 40% each to the first and last touch and shares the remaining 20% among the middle.
You build the journey as an ordered list of touchpoints, choose a conversion value, and the tool applies each model's weights to distribute that value across the channels involved. When a channel appears more than once, its credit is summed. The bars show each channel's share under each model at a glance.
The point is not to find the 'correct' model — there isn't one. It is to see how much the choice matters: a channel that looks worthless under last-click can be the top performer under first-click. Which model you adopt should follow from how your channels actually work together, not from a platform default.
Assumptions and limitations
- These are rule-based heuristic models. They assign credit by fixed position rules, not by measuring causal contribution — none of them prove a channel caused the conversion. For that you need incrementality testing or data-driven attribution.
- Real journeys are messier than an ordered list: view-through touches, cross-device paths, and touches your tracking never saw all change the true picture. This models the clean, observed path only.
- Time-decay here uses a fixed per-position half-life rather than real timestamps, and position-based uses the common 40/20/40 split. Analytics platforms let you tune these, so exact numbers vary by tool.
Frequently asked questions
What is multi-touch attribution?
Multi-touch attribution distributes credit for a conversion across every touchpoint in the customer journey, rather than giving it all to a single click. Different models — first, last, linear, time-decay, position-based — use different rules to split that credit, which is why the same journey can produce very different channel reports.
Which attribution model is best?
There is no universally best model; each encodes a different assumption about what matters. Last-click undervalues awareness channels, first-click undervalues closing channels, and the multi-touch models sit in between. Choose based on your goal — measuring demand creation versus demand capture — and ideally validate with incrementality tests rather than trusting any single rule.
Why do first-click and last-click disagree so much?
They credit opposite ends of the journey. A channel like paid social or display often starts journeys but rarely closes them, so it looks strong under first-click and nearly worthless under last-click. Seeing both side by side is the quickest way to spot channels your default report is systematically over- or under-valuing.
How does time-decay attribution work?
Time-decay gives more credit to touchpoints closer to the conversion, on the theory that recent interactions influenced the decision more. This playground halves the weight for each step further back from the close, then normalizes so the weights sum to the full conversion value. Real platforms use a time-based half-life instead of position.
Is my journey data stored anywhere?
No. The entire comparison is computed in your browser. Nothing about the journey you build or the value you enter is transmitted or stored.
More free tools like this, by email
I publish new calculators and explainers regularly. Get them when they land — no spam, unsubscribe anytime.