Guide

How to Measure Incrementality in Affiliate Marketing

Incrementality is the question every senior brand leader in partnerships eventually has to answer: would this sale have happened anyway? The answer decides how much of your affiliate spend is truly driving growth — and how much is subsidising demand you already had.

What incrementality actually means

Incrementality is the share of conversions attributed to a channel or partner that would not have occurred without their involvement. Last-click reports and multi-touch attribution both describe what happened; incrementality describes what would have happened otherwise. It is the only metric that separates partners who create demand from partners who intercept it.

Why it matters for senior brand leaders

• Budget defensibility. Finance and CMOs are increasingly asking channel leads to justify spend on true incremental revenue, not attributed revenue.
• Partner mix decisions. Loyalty, cashback, coupon, content, and creator partners have very different incremental profiles. You cannot pay them the same way.
• Commercial redesign. Tiered commissions, new-customer bonuses, and category-level rates only work if you know the incremental baseline you are paying against.

Four methods that actually work

1. Geo hold-out tests

Turn a partner or partner type off in a matched set of regions while leaving it on elsewhere. Compare total sales in the test vs control geos, not attributed sales. This is the cleanest read for channels with broad reach and works well for cashback, coupon, and top-of-funnel content partners.

2. Audience or user hold-outs

Randomly suppress a partner's placements or commissions for a defined user segment. Because assignment is randomised, the difference in conversion rate between exposed and held-out users is a direct read of incrementality. Best suited to placements you fully control (site extensions, on-site coupon boxes, retargeting-style creator campaigns).

3. Ghost bids and PSA tests

Serve the partner impression but strip the offer, or replace it with a public-service creative. Users are exposed to the placement without receiving the incentive. The delta in conversion rate isolates the value of the offer itself — critical when you suspect a partner is being paid for demand that would have converted at full price.

4. Difference-in-differences on partner launches

When a true experiment is not possible, compare the pre/post trend of the launched partner against a matched control cohort of similar partners. Weaker than a randomised test, but defensible when combined with a stable baseline period.

The pitfalls that break the read

• Testing on peak trading. Black Friday and launches distort baselines. Test in stable periods, then apply learnings to peak.
• Under-powered tests. Small brands rarely have the volume to detect a 5% lift in two weeks. Design for the effect size you can actually see.
• Single-partner conclusions. Incrementality is partner-specific, not channel-wide. One cashback partner's read does not license commission cuts across the whole programme.
• Ignoring long-term value. A partner with low first-order incrementality may still be introducing high-LTV customers. Measure both.

A practical 90-day starting point

1. Pick the two partners taking the largest share of commission and hypothesise their incremental profile.
2. Run a geo hold-out on the higher-volume of the two for a minimum of four stable weeks.
3. Read total revenue in test vs control — not attributed revenue — and calculate incremental ROAS.
4. Use the result to reshape commissions, new-customer bonuses, or placement scope for the next quarter.
5. Repeat with a different partner type. Build a living view of incrementality across the programme.

Where this conversation goes next

Measuring incrementality is a technical problem with a commercial answer. It reshapes how programmes are funded, how partners are paid, and how the channel is talked about at board level. It is one of the core themes senior brand leaders work through together at The Partnership Retreat.