1. Define the conversion
In a last click attribution program, define the conversion before advancing. Document the hypothesis, responsible owner, input evidence, accepted output, deadline and stop condition so the decision can be reproduced.
The output of this last click attribution step should be understandable to a reviewer who did not create the campaign or page. That discipline reduces hidden assumptions and improves future iteration.
2. List eligible touchpoints
In a last click attribution program, list eligible touchpoints before advancing. Document the hypothesis, responsible owner, input evidence, accepted output, deadline and stop condition so the decision can be reproduced.
The output of this last click attribution step should be understandable to a reviewer who did not create the campaign or page. That discipline reduces hidden assumptions and improves future iteration.
3. Audit identity coverage
In a last click attribution program, audit identity coverage before advancing. Document the hypothesis, responsible owner, input evidence, accepted output, deadline and stop condition so the decision can be reproduced.
The output of this last click attribution step should be understandable to a reviewer who did not create the campaign or page. That discipline reduces hidden assumptions and improves future iteration.
4. Choose the lookback window
In a last click attribution program, choose the lookback window before advancing. Document the hypothesis, responsible owner, input evidence, accepted output, deadline and stop condition so the decision can be reproduced.
The output of this last click attribution step should be understandable to a reviewer who did not create the campaign or page. That discipline reduces hidden assumptions and improves future iteration.
5. Select a model
In a last click attribution program, select a model before advancing. Document the hypothesis, responsible owner, input evidence, accepted output, deadline and stop condition so the decision can be reproduced.
The output of this last click attribution step should be understandable to a reviewer who did not create the campaign or page. That discipline reduces hidden assumptions and improves future iteration.
6. Document channel scope
In a last click attribution program, document channel scope before advancing. Document the hypothesis, responsible owner, input evidence, accepted output, deadline and stop condition so the decision can be reproduced.
The output of this last click attribution step should be understandable to a reviewer who did not create the campaign or page. That discipline reduces hidden assumptions and improves future iteration.
7. Compare model outputs
In a last click attribution program, compare model outputs before advancing. Document the hypothesis, responsible owner, input evidence, accepted output, deadline and stop condition so the decision can be reproduced.
The output of this last click attribution step should be understandable to a reviewer who did not create the campaign or page. That discipline reduces hidden assumptions and improves future iteration.
8. Validate with experiments
In a last click attribution program, validate with experiments before advancing. Document the hypothesis, responsible owner, input evidence, accepted output, deadline and stop condition so the decision can be reproduced.
The output of this last click attribution step should be understandable to a reviewer who did not create the campaign or page. That discipline reduces hidden assumptions and improves future iteration.
9. State limitations
In a last click attribution program, state limitations before advancing. Document the hypothesis, responsible owner, input evidence, accepted output, deadline and stop condition so the decision can be reproduced.
The output of this last click attribution step should be understandable to a reviewer who did not create the campaign or page. That discipline reduces hidden assumptions and improves future iteration.
10. Use the result carefully
In a last click attribution program, use the result carefully before advancing. Document the hypothesis, responsible owner, input evidence, accepted output, deadline and stop condition so the decision can be reproduced.
The output of this last click attribution step should be understandable to a reviewer who did not create the campaign or page. That discipline reduces hidden assumptions and improves future iteration.