What is a segment?

Kinds of segment

• Property segments — by what you know about people: country, plan, device, acquisition channel.
• Behavioural segments — by what they did: "purchased at least twice in 30 days", "used the export feature".
• Sequence segments — by the order of actions: "viewed pricing, then started a trial within a day".
• Value segments — by RFM score or spend tier.

The point of saving a segment is reuse: define "high-intent trial users" once, then look at their retention, funnels and revenue without rebuilding the rule every time.

Segments, cohorts and filters

• A filter narrows one report for now.
• A segment is a saved, named definition you can apply anywhere and compare with other segments.
• A cohort is a segment defined by when people did something — signed up in March — and is mostly used to track change over time.

Most segments are dynamic: membership updates as behaviour changes. Someone who stops buying leaves "active buyers" automatically.

Segments in TapCub

TapCub segments can combine behaviour, properties, sequences and RFM rules, for example "purchased two or more times in the last 30 days and never refunded". They are recomputed daily and can be used as a cohort in any analysis, as a filter on the people list, as the audience of a monitor and as an export.

Alongside segments, tags give each user one value — from a rule, a first or last event property, their top event, a static import or SQL — which is handy for labels such as "lifecycle stage" that need exactly one answer per person.

Common mistakes

• Segments too small to read. A segment of 40 people produces noisy rates.
• Overlapping segments compared as if separate. "Mobile users" and "paid search users" share people.
• Segmenting by outcome, then explaining the outcome. "People who bought have higher retention" is mostly true by definition.

Related guides: Customer Segmentation Analysis. See also user segmentation examples.

See also

• How do I create user segments?