01 · The number
Brand lift, without black-box claims.
Exposed vs. control. Confidence range shown. Methodology available for review.
SETUP
Lock the question
Audience, KPI, cells, and success threshold agreed before launch.
MEASURE
Exposed vs. control
Compare people who saw the campaign with people who did not.
REPORT
Decide what next
Lift, confidence, and the recommended next move in one report.
02 · Why EMV is not evidence
Reach is an input.
Lift is an outcome.
A campaign can reach millions and still move nothing. Brand lift measures what moved.
- Inputs
Reach · impressions · engagementUseful for delivery, not proof.
- Outputs
Awareness · consideration · preferenceMeasured against a held-out control.
- Decision
Scale · change · stopThe report says what to do next.
04 · Verified buyers
Attitudes are cheap. We measure them against actions.
Survey "intent" is one signal. We connect it 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 respondentsPanel-defined sample
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.