Incrementality: the truth about what your ads caused
Yesterday exposed that reported numbers are inflated. Today is the antidote and the most intellectually honest concept in performance marketing: incrementality - the only question that ultimately matters. Master this and you operate above almost everyone running ads.
1The one question that matters
Attribution asks "which ad gets credit?" Incrementality asks the deeper thing:
"How many of these conversions would NOT have happened without the ad?" If an ad gets credit for 100 sales but 80 of those people would have bought anyway, its INCREMENTAL contribution is 20. You paid for 100; you truly caused 20.
2Why it matters so much
The classic case (Day 18's correlation trap): you retarget add-to-cart users. They convert at a "fantastic" 10x ROAS. But many were going to complete the purchase regardless. The ad largely took credit for inevitable sales. Its true incremental ROAS might be a fraction of the reported number.
Conversely, upper-funnel prospecting that introduces brand-new customers often shows WORSE reported ROAS but higher INCREMENTALITY - those people genuinely wouldn't have found you otherwise. This is why optimizing purely to reported ROAS can quietly destroy a business: you shift budget to credit-claiming retargeting and starve the prospecting that drives new growth.
A vendor outside a station sells umbrellas and claims credit for every sale. But on a rainy day, most of those people would have bought from someone anyway - already wet, determined to buy. His INCREMENTAL contribution is only those who bought BECAUSE he was there and convenient. To find that number you'd compare a station with him to one without (the control). High sales ≠ high incrementality. The honest vendor measures the difference he made, not the receipts he can claim.
3How you actually measure it
You need an experiment with a control group - people who DON'T see the ad - to compare against:
- Conversion Lift / Geo tests - hold out a group or region; compare exposed vs held-out. The difference is the lift. Meta offers lift studies; geo experiments (some regions on, some off) are a robust DIY method - it needs comparable regions with enough order volume each to read a difference.
- Blackout test - turn off a campaign (or pause retargeting) and watch whether total sales actually fall. If they barely move, that spend wasn't very incremental. Crude but revealing. Run it for at least 1-2 full weeks (attribution lag plus day-of-week noise), compare total blended orders against a pre-period baseline or a matched region, and never run it across a promo or seasonal spike - a 2-day pause proves nothing.
- Blended + new-customer focus - blended ROAS and cost per NEW customer (Day 18) are far closer to incrementality than platform-reported, channel-credited ROAS.
Meta now offers an "incremental attribution" setting that models this for you - use it as a directional signal, but remember it is the platform estimating its own incrementality; independent holdout and geo tests remain the gold standard. (Setting labels move around; check the current name in your account.)
- Equating reported ROAS with value and over-funding retargeting that claims inevitable sales.
- Cutting prospecting because its reported ROAS looks worse, starving real growth.
- Never running a single holdout/geo test.
- Reporting platform ROAS to clients as incremental truth.
The apex differentiator: most agencies sell reported ROAS; the sophisticated few measure and optimize for incrementality and new-customer growth. Selling business impact, not reported ROAS, is how you win and keep serious clients.
Recap - 30 seconds
- Incrementality = how many conversions truly would NOT have happened without the ad. The real ROI.
- Reported credit overstates impact, especially retargeting; prospecting often has lower reported ROAS but higher incrementality.
- Optimizing purely to reported ROAS can starve the prospecting that drives real growth.
- Measure with holdout/Conversion Lift, geo tests, blackout tests, cost per NEW customer.
- Operating at the incrementality level - selling business impact - is the apex differentiator in this field.