Brand Lift Studies: What They Measure and What They Cost

What Brand Lift actually measures, and how it differs from a sales test

Every incrementality method covered elsewhere on this site — holdout tests, geo lift — measures sales. Brand Lift measures something upstream of a sale entirely: whether exposure to advertising changed what people think and feel about a brand. Amazon's own description covers brand recognition, attitudes, preferences, favorability, purchase intent and ad recall. The method is a survey, not a transaction record — Amazon runs it as a comparison between an audience that saw the ads and a matched control audience that didn't, asking both groups the same set of brand-perception questions and measuring the gap.

Eligibility and where it doesn't apply

Brand Lift is available for Amazon DSP campaigns and most Sponsored Ads formats — with one specific exception worth naming plainly: Sponsored Products is not eligible. That makes sense given what the product measures — Sponsored Products serves against an active search query from someone already close to a purchase decision, where a brand-perception survey has less to reveal than it would for an upper-funnel format built around awareness. Geographic availability differs by ad product too: DSP-eligible markets include the US, Canada, Germany, Spain, France, Mexico, Japan and the UK; Sponsored Ads eligibility for Brand Lift is currently limited to the US.

What it costs, and what Amazon doesn't publish

Brand Lift is offered at no direct cost, provided the campaign clears minimum spend and impression thresholds — but Amazon doesn't publish the specific numbers behind those thresholds on its public pages, directing advertisers to their account team or support for the details. That's worth stating plainly rather than guessing at a figure: this is a genuine gap in what's publicly documented, and any source quoting a specific minimum-spend number for Brand Lift eligibility should be checked against your own account rep before you plan a campaign around it.

Why survey-based measurement matters for exactly the inventory hardest to measure with sales data

Brand Lift's real value shows up most clearly for upper-funnel formats where sales-based measurement is structurally weak. In our own DSP book, streaming and connected-TV inventory returned the worst ROAS on the whole report, 0.77x, while carrying the highest new-to-brand rate of any placement, 56.3% — a pattern that's easy to misread as failure if you're only looking at attributed sales. A brand-perception survey run alongside that streaming spend answers a genuinely different, complementary question: did awareness and consideration actually move, independent of whether a same-session sale happened. For inventory this far upper-funnel, that's often a more honest read of whether the spend is working than a sales-attribution number that was never going to capture the effect cleanly in the first place.

The common mistake, including ours

The mistake is treating a positive Brand Lift result as proof of sales incrementality, when it's answering a related but genuinely different question — did perception move, not did the perception shift produce a sale. We've seen a client's team present a strong Brand Lift result as if it settled the ROI question for a streaming campaign, when the honest framing needed both numbers side by side: perception moved meaningfully, and the sales-incrementality question still needed its own separate test, because a lift in stated purchase intent on a survey doesn't automatically translate into a matching lift in actual purchases.

The reverse mistake happens too, less often but just as costly: dismissing a campaign that shows weak Brand Lift results as a failure without checking whether it was actually being asked to move perception at all, or whether it was a lower-funnel format being measured with the wrong tool entirely. Matching the measurement method to the funnel stage the campaign was actually built for is the check that catches both directions of this error.

What a survey-based method can catch that a sales test can't

There's a specific failure mode a sales-only measurement plan misses: a campaign that produces no meaningful lift in attributed sales during the test window, but does measurably shift how people feel about the brand, in ways that plausibly show up as a sale months later, well outside any reasonable test window. Brand Lift is one of the few tools available that can catch that kind of slow-burning effect directly, rather than waiting for it to eventually surface as a delayed sales pattern that's hard to trace back to its actual cause with any confidence.

How to use Brand Lift well

Treat Brand Lift as one input in a broader measurement plan for upper-funnel spend, not a replacement for sales-based incrementality testing. It's most useful paired with — not instead of — a holdout or geo-lift test on the same campaign: the survey tells you whether perception moved, the sales test tells you whether that perception shift (or the exposure itself) produced incremental purchases. Running both gives you a genuinely fuller picture of what an awareness-stage campaign is doing than either measurement alone.