Brand Lift · the keystone

Causal lift.
A number your CFO
will defend.

Exposed-vs-control brand-lift studies, pre-registered, audit-trailed. Not EMV, not reach, not "seemed to work". A methodology a statistician will sign off on — and a CFO memo finance will actually read.

METHOD: Exposed vs. control
SAMPLE: N ≥ 8,000 / market
CONFIDENCE: 95% CI default
AUDIT: Big-4 reviewed

01 · The number

Brand lift, the way your CFO wants it.

Exposed-vs-control. Pre-registered. 95% confidence interval. P-value. Panel composition on file. Re-runnable on demand.

PRE-REGISTER

Lock the hypothesis

Lift magnitude, cells, MDE, power, holdout — all locked before a single post goes live.

→ 72h setup

MEASURE

Exposed vs. control

Panels matched on the variables that matter. Day-zero reading preserved. Adstock modelled as it emerges.

→ 4-week read

AUDIT

Big-4 methodology on file

Panel composition, CI calculation, weighting scheme all open for review. Not a black box.

→ Open docs

02 · Why EMV is not evidence

"We reached 14M" is not the same as "14M more people know us".

EMV, impressions, reach — these are inputs. Brand lift is an outcome. A campaign can generate a billion impressions and zero awareness lift if the creative is wrong for the audience. We measure outcomes.

Inputs
Reach · EMV · impressionsUseful for pacing. Useless for proof. Every platform inflates these numbers with slightly different definitions.
Outputs
Aided awareness · consideration · preferenceWhat your board actually asked for. Measured against a held-out control that would have happened anyway.
Translation
From outputs → £ incrementalEvery lift study feeds a calibrated mix model. The awareness number becomes a margin number becomes a budget re-allocation.

03 · Methodology details

Fewer black boxes. More receipts.

PANEL

N = 8,000+ per market

Representative panels sourced through accredited providers. Weighted on age, gender, region, income.

DESIGN

Matched exposed/control

Panels matched on the 14 variables that predict baseline awareness. Residual imbalance reported.

ANALYSIS

Difference-in-differences

Standard error clustered at panel level. Multiple-comparison correction applied to secondary KPIs.

REPORT

One-page CFO memo

Auto-generated from the study. Plain English. Stat-sig marked. Ready to forward to finance without redaction.

04 · Verified buyers

Attitudes are cheap. We measure them against actions.

Most brand-lift studies ask strangers what they'd buy. Ours connects the panel to verified purchase behaviour — so the "intent" line on the page has a receipt behind it. We can route studies to people who have actually bought in the category, or overlay transaction data on top of survey responses to see which answers convert.

WHAT GOES IN

Survey respondentsN ≥ 8,000 / market

Verified buyers matched~62% of panel

Transaction history window24 months rolling

Category match precisionSKU-level

WHAT COMES OUT

Stated intent lift+14.7 pp

Actual repeat purchase · exposed+4.1 pp

Actual repeat purchase · control+0.6 pp

Intent → action conversion28%

Ask
Survey verified buyers of your categoryRoute the study to people who have actually bought yogurt, fragrance, trainers, whatever the category is. Real users answering, not opinionated strangers.
Overlay
Attach transaction history to every responseWhen a respondent says "I'd buy this", we already know what they've bought before. Their answer is weighted by their track record.
Prove
Watch intent become spendFollow the exposed cell forward. Did the awareness number turn into a basket number? The gap between what people say and what they do is the study.

Connection runs through our data partnership. Respondent-level PII is hashed and never leaves the processor. Your study reads aggregate outcomes — not individuals.

Causal lift.A number your CFOwill defend.