Glossary
Brand lift terms, defined.
The words that come up in brand lift reports, each explained in a sentence or two.
- 95% range
- The range of lifts consistent with the data at 95% confidence. If it crosses zero, the study cannot rule out no effect, and SpinProof reports the result as no measurable change.
- Absolute lift
- The gap between the exposed and control groups in percentage points. If 48% of exposed people recall the ad and 36% of the control do, the absolute lift is +12 points. SpinProof headlines absolute lift with a 95% range.
- Bogus respondent
- A survey participant who doesn’t answer sincerely, such as a fraudster, a click farm worker or an AI agent completing surveys for rewards. Bogus respondents can make rare answers look common and are hard to catch with standard checks.Read more →
- Brand lift
- The change in a brand metric caused by advertising, measured as the difference between people who saw the campaign (exposed) and comparable people who didn’t (control). Common metrics are awareness, ad recall, consideration, favourability and purchase intent.Read more →
- Cost per person lifted
- Media spend divided by persons lifted. It turns a percentage into money, so campaigns, placements and creatives can be compared on what each changed mind cost.Read more →
- Decoy ad
- An ad that never ran, shown alongside the real one in the survey. Anyone who claims to have seen it is over-claiming, so the decoy’s false-recall rate measures how far recognition answers can be trusted.
- Evidence File
- SpinProof’s audit trail for a study: method lock, sample flow, quality checks, agent run log, gate sign-offs and a results trace from every number to its source rows. It ships with the raw data and crosstabs in every study.Read more →
- Exposed group
- The people in a study who saw the campaign or ad. In a credible study, exposure is confirmed with recognition questions, a decoy ad and media data, not assumed from where people were surveyed.
- Headroom lift
- Absolute lift divided by the room left to grow above the baseline. It is a fairer reading for brands that most people already know, because it shows how much of the available headroom the campaign used.
- Holdout
- A group deliberately kept from seeing the advertising so it can act as a control. Who sets the holdout matters: if the platform that sold the media also chooses the holdout and reports the result, the measurement is not independent.
- Matched control group
- Comparable people from the same audience and time window who did not see the campaign. The only intended difference between exposed and control is exposure, so the gap between them can be read as the campaign’s effect.
- Method lock
- Fixing the study spec, survey version, primary KPI, sub-groups and analysis plan before fieldwork, then recording a cryptographic hash of them. If anything changes afterwards, the hash no longer matches.Read more →
- Minimum detectable lift
- The smallest lift a study can reliably detect, given its sample size, baseline and confidence level. At 300 people per cell and a 30% baseline it is about 10.5 points at 95% confidence and 80% power. Anything smaller will usually read as no measurable change.Read more →
- Persons lifted
- The number of people whose answer changed because of the campaign: absolute lift multiplied by the campaign’s reach. A +9.7 point lift on a reach of 2.35 million is about 227,000 persons lifted.Read more →
- Relative lift
- Absolute lift divided by the control group’s baseline. A move from 10% to 15% is +5 points but +50% relative. Relative lift flatters low baselines, so it should sit beside the absolute figure, not replace it.Read more →
- Sub-group analysis
- Reading results for a slice of the sample, such as one age band. Sub-groups are small and testing many of them produces false positives, so SpinProof only reports sub-groups fixed before fieldwork and adjusts their p-values with the Holm method.
Next step
Run a pilot study on your next campaign.
Pick a live or upcoming campaign in any channel. We agree the KPI with you, lock the plan before fieldwork, and you get a first read in week one, then the full report with its Evidence File.