Are you paying affiliates for sales you would have made anyway? A practical incrementality guide

Affiliate program reports tend to look good. The network dashboard shows a healthy return, partners are busy, and the commission line grows in step with revenue. The dashboard cannot tell you how many of those sales you would have made without the partner who got paid for them.

That gap has a name, incrementality. A sale is incremental when the partner caused it. A sale is not incremental when the shopper was already at your checkout, went looking for a discount code or a cashback rate, and a partner collected the last click on the way through. Your report shows both sales the same way, and you pay the same commission on each.

We run affiliate programs for D2C brands in the US, UK, Canada, Australia and Asia, and this is the first thing we check on any program we inherit. Below: where the leakage usually sits, three tests to find it, and how to redesign commission so the money follows the partners who grow the business.

What does incrementality mean in affiliate marketing?

Awin defines it as the additional value your affiliate activity delivers that would not have happened without it. That covers more than new customers. A partner that lifts basket size, converts a hesitant returning customer, or moves someone from a competitor to you is adding value too.

Most networks still pay on last click. Whoever touched the shopper last before purchase takes the full commission. Partners whose business sits right at the point of purchase, such as voucher sites, cashback portals and browser extensions, are built to be that last touch. That does not make them bad partners. It does mean their reported numbers look strongest, since last click measures the exact moment they are built to occupy.

The UK figures show the pattern. The Affiliate and Partner Marketing Association (APMA) found UK brands spent £1.8 billion on the channel in 2025 and tracked £20.7 billion in revenue, which APMA reports as a 15x return. Voucher partners posted the highest return of any type at 24x. That is an attributed return. It shows voucher partners were present at a lot of checkouts. Whether they caused those purchases is a separate question.

“

Awin defines it as the additional value your affiliate activity delivers that would not have happened without it.

Which affiliate partner types drive the most new demand?

Partners sit at different points in the buying journey, and where they sit predicts how incremental they tend to be. Awin groups them the same way: content, influencers, display and comparison partners early in the journey; brand partnerships, cashback and on-site tech at purchase; referral and loyalty after it.

The table below is our working ranking from running programs. Treat it as a typical pattern. A cashback partner that runs a dedicated email to a member base that has never heard of you can be highly incremental. A content site that only ranks for your brand name plus "discount code" is closer to a voucher site in practice.

Why do coupon and cashback partners claim so many last-click sales?

Picture a shopper who found you through a creator video, read a review, added to cart and then saw a discount code field at checkout. Many will open a new tab and search your brand name plus "code". The voucher site that ranks for that search gets the last click and the commission, and the creator who did the work gets nothing.

Browser extensions take this further by sitting inside checkout itself. Google tightened the Chrome Web Store affiliate ads policy in 2025 after public criticism of how some coupon extensions handled affiliate links. Since enforcement began on 10 June 2025, an extension may only apply an affiliate link, code or cookie when the user takes an action and gets a direct benefit, such as a working discount. An extension that finds no coupon must not insert an affiliate link. That closes the worst behaviour. These partners still get paid for being present at the end of a journey someone else started.

Cashback is more mixed. Members do browse cashback apps to choose where to shop, especially in peak season, and that can be real demand. Members also learn to detour through the app before buying something they had already decided on, and your order data can show which pattern you have.

None of this means you should drop voucher or cashback partners. They convert hesitant shoppers, they matter to price-sensitive buyers, and at peak a cashback homepage slot can reach people who would never have found you. Pay them for what they add, and stop paying for every order they happen to touch.

How do you test whether an affiliate partner is incremental?

No single test gives a perfect answer, so we use three, chosen by program size and risk. Awin argues against blanket blackout tests because they can hand market share to competitors and damage partner relationships, and we agree. Pause narrowly, for a fixed period, with the partner told in advance where the contract requires it.

• Partner pause (holdout). Switch off one partner or one partner type for two to four weeks and compare total store revenue, not just affiliate revenue, against the same weeks before and a matched control period. If store revenue barely moves, most of that partner's sales were going to happen anyway. Run it outside peak season and never on your biggest partner first.
• Geo split. If you sell in several markets, run the pause or a commission change in one region and keep another as control. A brand selling in the UK and Australia can test a voucher pause in one and compare the change in total revenue against the other. This needs enough volume in each market to read a difference.
• Coupon-code isolation and new-versus-returning. Give each partner its own unique code, stop public codes leaking onto voucher sites, and pass customer status from Shopify into your network so every sale is tagged new or returning. A partner whose sales are mostly returning customers using a code they could have found anywhere is a retention cost, and you should price it as one.

Is there evidence that affiliates can be incremental?

Yes. Read this before you cut anything. CJ ran a test-and-control study across 20.2 million shoppers and 5.5 million transactions for large household-name retailers, comparing shoppers who interacted with affiliates against similar shoppers who did not, over a 30-day window. Affiliate shoppers converted at 5.8% against 4.0%, spent 29% more per customer, and produced 88% more revenue per shopper. For new customers, conversion was 140% higher.

Two caveats. The study is from 2018 and was run by a network, so treat it as a strong signal rather than a benchmark for your brand. It also measured the channel as a whole, and that average hides a wide spread between partners. You still need to test them one by one.

How should you structure affiliate commission for incrementality?

Once you know which partners grow the business, the commission plan should pay them more and pay the rest less. Most networks now support this without custom code. Awin, for example, accepts a NEW or RETURNING customer parameter on each sale and lets you set different rates for each in its commission rules. impact.com and other platforms support similar customer-status rules.

Our usual building blocks are below. None of them require cutting a partner outright, which keeps relationships intact while the incentives change.

• New-customer rates. Pay a premium for first orders and a lower rate for returning customers. A common starting point is roughly double the returning rate for new customers, then adjust from test results.
• Category rates. Pay more on full-margin lines and less on clearance, bundles or products that already sell on their own. This stops partners pushing your lowest-margin items at your highest rate.
• Content and placement bonuses. Pay flat fees or bonuses for reviews, gift-guide inclusions and creator videos that start journeys, rather than relying on last-click commission alone to reward discovery.
• Code controls. Give voucher partners partner-specific codes with a defined offer, and decline sales where a code was used that the partner was never issued.
• Tiered rates for growth. Raise a partner's rate once they pass a threshold of new customers in a period, so the reward follows acquisition rather than raw order volume.

How long should your affiliate validation window be?

Commission is meant to be paid on confirmed sales, so the validation window, the period before you approve or decline each transaction, is where you enforce your rules. Returns, cancellations, fraud and unissued codes are all declined here.

We use windows of up to 90 days. The right length matches your returns policy plus a buffer for refunds to clear. A fashion brand with a 30-day returns policy needs longer than 30 days, because returns arrive late and refunds take time to process. Keep the window consistent, publish it in your program terms, and approve on time. In our experience partners accept firm rules far more readily than slow or unpredictable payment.

Use the window to check more than returns. Look for orders where the code used was never issued to that partner, sales from the same customer split across several partners, and spikes from a partner with no matching promotion. Each one is a sign that commission is drifting away from the work you meant to reward, and the validation window is the cheapest place to fix it.

Which affiliate networks matter in each market?

The tests and commission rules above work on any network, but the partner mix and the network you run on differ by market. Customer-status tracking, coupon controls and commission rules vary by platform, so check what yours supports before you design the plan.

1. 1.APMA State of the Affiliate Nation 2025, via Martech Record
2. 2.InPublishing: Affiliate and partner marketing spend grows to £1.8 billion in 2025
3. 3.CJ: A Study in Affiliate Incrementality
4. 4.Awin: Understanding incrementality
5. 5.Awin: Customer acquisition tracking
6. 6.Awin: Commission groups
7. 7.Chrome for Developers: Chrome Web Store affiliate ads policy update
8. 8.Awin: ShareASale to Awin upgrade
9. 9.Involve Asia
10. 10.Accesstrade launches in Malaysia (Digital News Asia)
11. 11.A8.net
12. 12.vCommission