- An attribution model allocates credit under stated rules.
- Direct-traffic treatment and eligible touchpoints can change the answer.
- A credit assignment is not proof of incremental impact.
Use the same path before comparing models
Consider an illustrative customer who arrives through Google search, later clicks a newsletter, and finally returns directly before a $100 conversion. Three rules can allocate the same $100 differently without any arithmetic error. The rule decides which eligible touchpoints receive credit.
First-touch gives the full amount to the earliest included touchpoint. Last-touch gives it to the latest included touchpoint. Linear attribution divides it equally across included touchpoints. These are simple educational definitions; an actual platform may exclude certain visits, use a different identity boundary, or apply a lookback window.
Work through the credit allocation
Under the simulator’s rules, Google search receives $100 in first-touch, Direct receives $100 in last-touch, and each of the three touchpoints receives approximately one third in linear. Because currency has cents, our example assigns $33.34 to the earliest touchpoint and $33.33 to each remaining touchpoint.
All columns still sum to $100. A second newsletter touchpoint would earn another share under linear allocation because this model divides per touchpoint, then groups by channel. That makes repeated exposure relevant to the calculation without proving repeated exposure caused more value.
| Channel | First-touch | Last-touch | Linear |
|---|---|---|---|
| Google search | $100.00 | $0.00 | $33.34 |
| Newsletter | $0.00 | $0.00 | $33.33 |
| Direct | $0.00 | $100.00 | $33.33 |
| Total | $100.00 | $100.00 | $100.00 |
Check Direct before comparing with a platform
The free simulator treats Direct as an ordinary touchpoint in all three models. That is a teaching choice, not an attempt to recreate GA4. Google’s current attribution documentation describes different available models and generally excludes Direct from credit when another eligible channel exists.
Before reconciling reports, write down the platform’s selected model, channel eligibility, direct-traffic rule, conversion definition, date basis, time window, timezone, and currency. Also check which visits can actually be connected. A model applied to incomplete observations may be internally consistent while leaving the acquisition story incomplete.
Ask the question the model can answer
First-touch can help describe the first included discovery source. Last-touch can help describe the final included return. Linear can illustrate a shared-credit convention. None of these simple rules tells you what would have happened without a particular campaign.
For example, removing a newsletter does not necessarily remove the customers credited to it. They may have returned another way, or never returned at all. Assessing incremental impact requires stronger evidence and an appropriate comparison, rather than treating an allocation column as an experiment.
Keep the simulator separate from measured revenue
Use the tool to explain an attribution disagreement to a colleague. Enter an ordered fictional path, compare the allocations, and export a CSV. The entries stay in your browser. Changing the currency changes the label; it does not perform an exchange conversion.
VisitLift’s installed reports follow the product’s own attribution and verified-Checkout definitions. The simulator does not add attribution models to your workspace or read actual payments. When using payment reports, compare the same covered population and keep unmatched revenue explicitly unattributed.