Deciding when an affiliate placement is not incremental

Judge a placement not meaningfully incremental for a stated outcome only when a credible comparison and its uncertainty rule out a gain large enough to matter under the tested conditions.

A credited order, late click or low new-customer rate warrants investigation, not a verdict. A placement may add some demand yet cost more than that gain is worth; call that commercially insufficient, not no effect.

Set the decision boundary

Name the partner, page or slot, audience, offer and period. Choose the outcome: extra approved orders, first-time customers, contribution or another business measure. Set the smallest gain worth pursuing before seeing the result. Record commission, placement fees and funded discounts separately so their treatment in the outcome is clear.

A checkout voucher slot might be assessed for additional contribution from its exposed audience. A product guide might need a longer observation period to assess first-time-customer orders. These are possible evaluation choices, not findings about either format.

Key Metrics for Evaluating Affiliate Placement Incrementality

Check execution, then evidence

First confirm that the placement ran, the offer worked and eligible customers could buy the product. Reconcile credited orders to store records and distinguish pending from refunded orders. A broken destination or unavailable item is an execution problem; it does not establish the value of a working placement.

Next examine customer status, timing of recorded touches, overlap with paid search and where the offer appeared. These indicators can form a hypothesis but cannot show the outcome without the placement. A first-time buyer may already have intended to buy; a returning buyer might still make an additional purchase because of the placement.

Use the strongest feasible comparison. A planned holdout can compare total store outcomes for eligible groups with and without access. A geographic or staged comparison may provide weaker, qualified evidence if its groups and concurrent changes are examined. Treat a simple before-and-after change as directional when other conditions moved.

A range entirely below the commercial threshold supports a stop decision. It does not prove the effect was exactly zero. To use the stronger phrase “not incremental”, state the outcome and how the evidence bounds any possible positive effect.

Act on the finding

If the tested placement falls short, record its version before changing the audience, position, offer or product. A different version may perform differently. If it adds demand but costs too much, revised terms may be worth discussing under the partner agreement.

Awin Classic’s Publisher Incrementality report compares partner-involved baskets and conversion rates with site averages. It is available only on Awin Classic and for Awin Advanced customers. Those indicators may help choose what to test, but they do not provide a controlled no-placement result.

The decision record should identify the placement, outcome, comparison, uncertainty, costs and action. Apply the conclusion only to the conditions the evidence covers.

Pros and Cons of Using Awin Classic’s Publisher Incrementality Report