Always a comparison
The term is used for every variable an audience can be cut by: age, gender, region, interest, media channel, brand. It is always a statement about a comparison, never about a raw share.
Example (illustrative): a cycling audience that is 22% women overindexes on women if the baseline is 14%, and underindexes if the baseline is 31%. The 22% on its own says nothing.
Why underindexing is often more actionable
- It shows absence. A brand that underindexes in a region where its category is strong usually has a distribution or awareness problem, not an audience problem.
- It sizes the headroom. The gap between an underindexing group and the baseline is volume a campaign could realistically win.
- It protects against preaching to the converted. Budget tends to flow to overindexing groups because they look like success in a report, even where the additional return is low.
Common mistakes
- Points vs index. A move from 2% to 4% is +2 percentage points but an index of 2.0x: small in volume, large in lean.
- Cutting until something overindexes. Enough slices will always produce an extreme value somewhere. Decide the cuts before you look at the numbers.
- Ignoring base size. Publish the absolute number next to every index. A lean computed on a few hundred people is noise.
How Rascasse uses it
In Rascasse, every audience shows where it over- and underindexes: by region and city, by demographics, by interests, brands and media, each against the population of the same country. Regions are compared down to postal code where available.
Related use cases: Audience Profiling, Local Targeting & OOH, Media Planning.
