Rascasse› Glossary› Affinity Index

Audience Intelligence Glossary

What Is an Affinity Index?

An affinity index expresses how much more — or less — common a behavior is in one group than in a reference population. It is the arithmetic behind almost every statement of the form "this audience is twice as likely to…".

The index is deliberately simple. Its usefulness comes from the fact that it removes size from the comparison: a segment of 40,000 people and a segment of 4 million can be compared on the same scale, because both are measured against the same baseline.

The formula

Affinity index = (share of the segment showing the behavior) ÷ (share of the baseline population showing the behavior)

Example: 9% of a team's followers search for an energy drink; 3% of the national population does → index 3.0x (or 300)

Two conventions exist for the same number, and confusing them is the most common reporting error in audience decks:

  • Multiplier notation — 1.0x is the baseline, 2.0x is twice as likely. Common in digital and social analytics.
  • Index-100 notation — 100 is the baseline, 200 is twice as likely. Standard in classical market research and media planning (MRI, TGI, AGF).
  • They are the same value scaled by 100. An index of 120 and an affinity of 1.2x describe an identical finding.

Choosing the baseline decides the answer

An index is only as meaningful as the population in the denominator. The same segment can show an index of 3.0x against the national population and 1.1x against people who already buy the category. Both are correct; they answer different questions.

  • National baseline answers "is this audience unusual compared to everyone?" — useful for positioning and for finding untapped groups.
  • Category baseline answers "is this audience unusual compared to our existing buyers?" — useful for deciding where incremental budget goes.
  • A report that states an index without naming its baseline cannot be checked, and should be sent back.

When a high index is misleading

  • Small denominators. If only a few hundred people sit in the segment, a handful of individuals can move the index from 1.0x to 4.0x. Always read the index together with the absolute size.
  • Correlated definitions. Measuring the affinity of "football fans" for "football boots" produces a large number that contains no information.
  • Index without reach. A 6.0x index on an audience of 12,000 people is usually worth less than a 1.4x index on an audience of 4 million. Index tells you concentration; it says nothing about scale.

How Rascasse calculates it

Rascasse derives indices from anonymized search and social behavior rather than from survey responses, which means the baseline is an observed population rather than a weighted panel. Every affinity shown on a profile names its reference group, and every profile is computed per country — an entity has different indices in Germany and the United States, and the German number is never used as a stand-in for the American one.

Live example

Compare the affinity views for Adidas in the United States and Borussia Dortmund in Germany — same logic, different reference populations.

Related terms