17 Nov, 2026
Written by
Claudia Argomenti
Few subjects come up in performance marketing as often as attribution and incrementality. They surface in quarterly reviews and in nearly every discussion about commission terms, and people often use the two words as though they were interchangeable.
They aren't.
Attribution is an accounting exercise. It decides which touchpoints get credit for a conversion, using whatever methodology a business chooses to apply. Incrementality asks a harder question: what would have happened without that marketing activity? A partner can sit at the top of an attribution report and still drive very little incremental revenue if the customers it claims were already on their way to buy.
That distinction was at the heart of a recent panel discussion with Kurt Kirby of QVC, Calum Lewis of Dell Technologies and Nick Leyland of Cardlytics, moderated by Claudia Argomenti, Client Development Director at CJ. Their experience points to a bigger opportunity for the partnership industry: incrementality can prove performance, and it can also help marketers make better decisions about partners, campaigns and future activity.
According to the panellists, the trouble often starts before any test is run.
Incrementality can mean different things to different organisations, and sometimes to different teams within the same organisation. QVC's experience shows how much it helps to agree at the outset on what counts as an incremental order or outcome, which business question needs answering and what success should look like.
The practical barriers matter too. Sound testing needs the right data, a credible methodology, enough resources and internal backing, and those capabilities vary between advertisers, publishers and networks.
Cardlytics adds a further point. Before deciding how to assess incrementality, it is essential to understand how the advertiser already defines and measures success, including how it measures comparable channels. With first party transaction data on 24 million UK bank customers, Cardlytics can usually replicate that approach, so results are directly comparable.
In CJ's work with clients, tests go better when everyone agrees on four questions before they start:
Settling those first makes the next question far more meaningful.
Once the business question is clear, marketers can look at what their existing reporting can and can't answer.
At Dell, attribution was showing who received credit, though it couldn't show who created value. That prompted a different question: what would happen if we stopped this activity?
At QVC, questions about causality have made incrementality testing a growing part of how the company evaluates performance.
Cardlytics comes at the problem from the publisher side with an equally plain question: would those customers have spent anyway? Its test-versus-control approach examines that question across new, lapsed and existing customers, including incremental spend per visit and frequency.
CJ's view lines up with all three. Attribution tells us where credit went. Incrementality testing compares what happened with an estimate of what would have happened without the activity, which tells us what additional outcome it produced. Used together, the two give a far more complete picture.
Admittedly, this is easier said than done without the right tools and technology. This is why CJ has invested heavily to ensure we have a suite of solutions to better help our clients understand their incrementality. Our incrementality and measurement framework helps us understand incrementality at the channel and partner level. Additionally, our Program & Partner optimisation solution supports commission rate elasticity testing and effective campaign planning.
One of the most useful things a test can do is show that an assumption was wrong.
Dell found that some partners who looked relatively inefficient in attribution reporting were generating significant incremental value. Others received substantial attribution credit without delivering a matching level of net-new impact.
QVC's experience adds that one test rarely settles the matter. Tests confirm and disprove assumptions, and they also raise new hypotheses as partners, activity and performance change over time. Marketers need to be prepared for evidence that upends partner rankings, budget decisions or long-held views about where value comes from.
That makes incrementality a repeating cycle:
Hypothesis → test → evidence → decision → new hypothesis
The panel also discussed how CJ's Affiliate Customer Journey and Cross Channel Journey reports give marketers critical insight into incrementality. Both provide unique visibility into the customer journey across partners and channels and help identify the true value affiliates drive across different touchpoints.
One implication is easy to miss. Incrementality is a publisher responsibility as much as an advertiser one.
Kurt Kirby says QVC increasingly leans towards partners that can support incrementality testing, understand QVC's business and hypotheses, and help demonstrate incremental value.
Cardlytics goes further, using test findings to shape its recommendations on targeting, cashback levels, minimum-spend requirements (set above a customer's typical basket) and offer duration. Nick Leyland believes publishers should be proactive even when they can't run sophisticated tests on their own and should work with advertisers on approaches such as holdouts or pre/during/post analysis. By being transparent regarding their method and suggested approach, publishers can become trusted growth partners and not just a line item.
CJ works with many publishers to help them become real measurement and research partners to the advertisers they serve. Publishers don't need to build their own incrementality methodology, but they should be ready to:
Publishers that do this get a direct hand in demonstrating the value they create.
At Dell, the conversation moved towards judging partners on the value they created as well as the credit they received, and testing informed how different types of partner activity figured in investment decisions. At QVC, incrementality continues to shape how partners and investment are evaluated. At Cardlytics, evidence feeds straight back into the design and optimisation of future activity.
Every incrementality test should end with one question: what should we do differently because of what we learned? The answer might be to:
If a result doesn't inform a decision, the measurement exercise has delivered very little.
CJ has also recognised the need to support brands as they develop a holistic view of all their channel investments, including affiliate. For example, our data science team is working with multiple advertisers and their Media Mix Modelling (MMM) service providers to ensure affiliate is accurately represented. Measuring affiliate within MMM models is challenging because MMM looks for correlations between changes in channel spend and their impact on revenue. Because affiliate spend tends to be more modest and the spend pattern is often “always on”.
QVC, Dell and Cardlytics all showed that effective testing depends on advertisers, publishers and measurement partners working from the same questions. Waiting for perfect measurement only delays the learning. Pick a clear business question, gather enough evidence to make a better decision, learn from the results and use what you learn to sharpen the next test.
Want to go further? Watch Dan's interview on Hello Partner for more on measuring true partner value.
You sell or promote products and services and work with websites and creators, paying commissions for sales, leads or traffic.
You've built an audience or following and want to earn commissions by promoting brands and products you believe in.
Share