Incrementality

What it is

Incrementality is a causal measurement discipline. Rather than asking which touchpoint preceded a conversion, it asks what would have happened if the activity had not run at all. The gap between the two outcomes, the treated group and the control, is the incremental result.

Why it matters

Attribution models credit channels that were simply present near a conversion, which flatters branded search, retargeting and anything that sits late in the journey. Incrementality tells you whether budget is creating demand or harvesting demand you already had. For AI driven discovery, where models and assistants increasingly sit between the brand and the buyer and click level tracking is thin, causal testing becomes one of the few reliable ways to judge whether an investment is working.

How it works

The common methods are geo holdout tests, where matched regions are switched off or on; randomised user level experiments through ad platform lift tools; and time based switchbacks that alternate activity on and off in short cycles. Analysts match test and control on baseline behaviour, run long enough to cover the typical purchase cycle, then compare outcomes and report the lift with a confidence interval. Results are usually translated into incremental cost per acquisition or incremental return, which is then compared with the blended figures reporting tools produce.

When it applies

Use it when a channel's reported performance looks implausibly strong, when you are deciding whether to scale or cut a budget line, or when tracking signal has degraded and platform reported conversions can no longer be trusted. It applies to any activity with enough volume to detect a difference above noise.

Examples

• Switching off brand search ads in half of a country's regions for four weeks to see how much of the traffic and revenue is recaptured by organic listings.
• Running a randomised holdout on a retargeting audience, where a share of users are deliberately excluded, to test whether the ads create sales or follow them.
• Pausing a content and digital PR programme in one market while continuing in a matched market, then comparing branded search volume and direct sign ups over the following quarter.

How it is measured

• Incremental lift, expressed as the percentage difference in conversions between the test and control groups.
• Incremental cost per acquisition, calculated as spend divided by incremental conversions rather than all reported conversions.
• Incremental return on ad spend, comparing incremental revenue with the cost of the activity.
• Statistical confidence and the width of the confidence interval around the lift estimate, which indicates whether the test had enough power to trust the result.