Incrementality in Marketing: Measuring the Real Impact of Campaigns

Attribution metrics have been a standard in digital marketing, but they have a key limitation: they report on what happens after an interaction, without revealing if that conversion would have happened anyway. This is where incrementality comes into play, allowing one to distinguish the real effect of a marketing action. It's relevant in affiliate and partner programs, where many channels claim attributions that do not necessarily reflect causation.

What Does Incrementality Really Measure?

Incrementality answers the question: "How many additional conversions, customers, or revenues did this campaign generate that wouldn’t have happened without it?" In other words, it measures the extra outcome caused by a marketing action, eliminating the user’s native or inherent demand. This allows one to know if an affiliate genuinely generated demand or merely captured a pre-decided purchase.

For instance, if a brand attributes 10,000 sales to its affiliate program but a test shows that only 1,000 of those sales were additional compared to a control group, the real contribution of the channel is 1,000 incremental conversions.

Differences Between Attribution and Incrementality

Attribution assigns credit, but incrementality measures causality. A last-click model assigns the sale to the last channel, while a multi-touch model distributes credit among several. Neither guarantees that the marketing activity caused the conversion.

Google emphasizes this difference in its documentation on Conversion Lift: attributed conversions are a result of configured rules, while incremental conversions come from comparing an exposed group with a control group. Attribution shows participation; incrementality shows cause and effect.

Why It’s Key to Combine Attribution and Incrementality in Affiliates and Partners

It is not prudent to view attribution and incrementality as rivals. They work better together. Attribution manages the day-to-day: user journeys, partner activity, and immediate performance. Incrementality adds a causal layer that can confirm or challenge the attribution report.

A typical result: a content partner shows $500,000 in attributed revenue, but an incremental test reveals that their audience does convert more against an unexposed group. Thus, the brand knows if the reported results reflect true effect.

Incrementality, the New Standard Amid Budget Pressure

Today, users pass through multiple touchpoints — organic search, social media, affiliate blogs, retargeting, marketplaces — before converting. Multiple platforms claim attribution over the same conversion. In the face of the demand to justify every dollar spent, incrementality shifts the conversation from "who receives credit" to "what causes marketing proved."

According to a Nielsen analysis of campaigns on Pinterest in the CPG sector in Canada (2026), 80% of the campaigns generated statistically significant incremental sales. This type of approach allows:

• More accurately distribute budgets
• Identify which partners drive real growth
• Reduce overlaps/cannibalization among channels
• Better segment the customer
• Improve commission structure
• Timely detect declining returns

How to Apply Incrementality in Affiliate and Partner Marketing

In affiliate marketing, partners impact customers at different stages along the buying journey. While a publisher may introduce the brand, an influencer generates consideration, a comparator facilitates decision-making, and a cashback program may only close a pre-decided sale.

It’s not enough to only look at attributed conversions. Useful metrics include:

• Incremental conversion ratio
• Incremental revenues
• Incremental new customers
• Incremental CPA and iROAS
• Average ticket, new customer rate, customer lifetime value
• Reactivation rate and commissions on incremental revenues

For example, Partner A generates 10,000 attributed orders but little real increment, while Partner B achieves fewer attributed orders but greater incremental growth of new customers. Viewed through incrementality, Partner B may be the most effective growth engine.

How to Measure Incrementality: Main Methods

There is no one-size-fits-all test. Choosing the technique depends on the control available over exposure to the campaign.

Difference Between iROAS and Classic ROAS: iROAS calculates the revenue actually caused by investment, eliminating conversions not attributable to the activity.

Most Used Incremental Testing Methods

1. Random Holdout Test: divides the audience into treated group (exposed to the campaign) and control group (not exposed); the variation in results estimates the incremental impact. Example: by withholding an affiliate promotion from 10% of users, if the conversion of the exposed group is 3.2% and the control is 2.8%, the incremental lift is 0.4 percentage points.
2. Geographic Test: uses regions or cities as treatment/control groups instead of individual users. This helps when it’s not possible to control individual exposure. Example: a partner promotion in 10 cities while another 10 similar ones do not receive it.
3. Scale Test: examines the effect of increasing investment. If $20K generates $60K (iROAS 3.0x) and raising the spending to $60K reaches only $105K (iROAS 1.75x), the channel remains profitable but shows lower lift per additional dollar.

Risks and Conditions for Valid Incrementality Tests

Even robust tests can lead to errors if certain practical points are not considered:

• Contamination of the Control Group: people can be exposed through other channels (search, email, word of mouth), distorting the result. It’s crucial to map potential leaks before launching the test.
• Seasonality: holiday periods, competitor promotions, or cycle changes can alter baseline demand and create false increments.
• Sample Size: small groups can yield apparent but unreliable differences. Planning the sample size before the test is what provides statistical validity.

Framework for Performance Marketing Decision-Making

Incrementality does not replace attribution; it complements it. The optimal approach is to combine attribution, experimentation, and business metrics. Some useful rules are:

• High attributed performance + high incremental: scale up
• High attributed + low incremental: investigate cannibalization
• Low attributed + high incremental: explore underattribution and increase investment
• Low attributed + low incremental: reconsider the channel or partner

An example: a content partner contributes less attributed revenue than a coupon partner but generates greater incremental growth of new customers. The advertiser can increase investment in content and rethink the role of the coupon channel.

"Attribution tells you where the credit went; incrementality tells you what caused the marketing." — Google Conversion Lift Document

The goal is not to find the channel with the highest attribution but to know which investment generates measurable additional growth. Repeating experiments improves confidence in decisions regarding budgets and optimization. In affiliate marketing, incrementality installs itself as the tool that allows differentiation of partners and channels that truly add additional value.