TapCub is live — analytics, insights and live chat, free on one platform
Growth & Marketing

Audience Segmentation Tools: Ten Criteria to Compare Before You Choose

Ten criteria with a pass test each, a weighted comparison matrix, five segments built from events with sizes and intervals, a one-hour trial and the privacy checks for profiling.

Sophia Bennett Published Nov 21, 2026 11 min read Growth & Marketing
Sample data: four segments defined by events, a window and a count. Each has enough members for a narrow 95% interval, such as 9.1% to 13.2% for cart abandoners.Sample data

Audience segmentation tools define a group of people by a rule, keep the group current as behaviour changes and let you analyse or pass it on. Compare them on ten criteria: event rules with did and did-not, rolling windows, update mode, identity, size preview, reuse, export, privacy, limits and explainability. Below: a pass test for each, a weighted matrix, five segments built from events and a one-hour trial.

What you will learn

  • Criteria 1 to 3 (event rules, time windows, update mode) decide whether behaviour-based segments are possible at all
  • Score options with weights; the ranking can change when your priorities change
  • Check segment size: 900 people at 11% gives 9.1% to 13.2%, while 4 of 40 gives 4.0% to 23.1%

What do audience segmentation tools actually do?

An audience segmentation tool does three jobs: it defines a group of people by a rule, it keeps the group current as people's behaviour changes, and it makes the group usable, either by analysing it or by passing it to another system. Tools differ mostly in how well they do each job when the rule is built from behaviour, such as "added to cart in the last 7 days and did not buy", rather than from a list of attributes such as country or plan.

The idea of dividing a market into groups with different needs goes back to Wendell Smith's 1956 paper on product differentiation and market segmentation. What changed since then is the data: with event tracking you can define a segment from what people did last week instead of who they are. That is behavioural segmentation, and it is the lens for this guide; for the definition and examples see behavioral segmentation, and for the method see how to segment your audience and audience segmentation.

This guide gives you ten criteria to check, a weighted matrix to compare options without naming products, five example segments built from events with sample numbers, a one-hour test, and the privacy checks that apply when segments are used for marketing.

Which ten criteria should you check?

#CriterionQuestion to askPass test
1Event rulesCan a rule say "did" and "did not do" an event, with a count?Build "did purchase 2+ times and did not refund"
2Time windowsIs the window rolling (last 30 days) rather than fixed dates?Same rule gives different members next week
3Update modeDaily refresh, real-time entry, static list: which are supported?Do an event, time how long until membership changes
4IdentityDoes it handle anonymous visitors, known people and accounts?See whether anonymous visitors are members
5Preview and trendIs member count shown before saving, and tracked over time?Count matches a query you run separately
6ReuseCan the same segment filter funnels, retention and reports?Apply it to a funnel in one click
7Export and activationCan membership go to another system (file, API, webhook)?Receive a change notification
8PrivacyDoes it respect consent, retention and data-subject requests?Delete a person, see them leave
9LimitsMaximum members, number of segments, refresh frequency?Read the plan limits before you build
10ExplainabilityCan you see why a given person is in the segment?Open one member and read the match

Criteria 1 to 3 decide whether you can build behaviour-based segments at all. If a tool cannot say "did not do", you cannot build lapsed-buyer or abandoned-cart segments, which are the ones most often used.

How do you compare options with a weighted matrix?

Weight each criterion by how much it matters to you (3 = decisive, 2 = important, 1 = nice to have), score each option 0 to 2 and multiply. Sample data: three hypothetical tools, not products.

#CriterionWeightTool ATool BTool C
1Event rules3221
2Time windows3212
3Update mode2210
4Identity2122
5Preview and trend2210
6Reuse2221
7Export and activation1122
8Privacy1212
9Limits1121
10Explainability1211
Weighted total (max 36)322721

Tool A scores 32 of 36 (89%), B scores 27 (75%) and C scores 21 (58%). Tool B wins on export and limits but loses where the weights are highest. Change the weights to your situation and the order can change: if you mainly need to push audiences to an email system, give criterion 7 a weight of 3.

CriterionPass testWeight
Event rules with did and did-notBuild did purchase 2+ times and did not refund3
Rolling time windowsSame rule gives different members next week3
Update modeTime membership change after a test event2
IdentitySee whether anonymous visitors are members2
Size preview and trendCount matches a separate query2
Reuse in funnels and retentionApply segment in one click2

Remaining four criteria (export, privacy, limits, explainability) carry weight 1 in the sample matrix.

What do segments built from events look like?

Sample data: a product with 20,000 identified users and these five rules.

SegmentRule (events)MembersShare of base30-day conversion95% interval
Power usersActive on 8+ days in the last 301,8009.0%22% (396)20.1% to 24.0%
Cart abandonersAdded to cart in 7 days, did not buy9004.5%11% (99)9.1% to 13.2%
Pricing viewersViewed pricing in 14 days, did not start a trial1,5007.5%6% (90)4.9% to 7.3%
Lapsed buyersBought in the last 180 days, not in the last 602,40012.0%4% (96)3.3% to 4.9%
Everyone elseNone of the above13,40067.0%not shown

Three things to read from the table:

  1. Size before ambition. The smallest segment has 900 people. At 11% conversion its interval is 9.1% to 13.2%, narrow enough to act on. A segment of 40 people with 4 conversions (10%) would run from 4.0% to 23.1%, and you should not build a campaign on it.
  2. Rules are events plus a window plus a count. Every row has an event ("did" or "did not"), a window (7, 14, 30, 60, 180 days) and, for power users, a number (8+ days).
  3. Segments overlap. A power user can also be a pricing viewer. Check overlap before adding member counts together; the 33% of the base in the first four rows is an upper bound.
app.tapcub.com

30-day conversion by event-based segment

sample base of 20,000
Power users (1,800)22%
Cart abandoners (900)11%
Pricing viewers (1,500)6%
Lapsed buyers (2,400)4%
Sample data
Sample data: four segments defined by events, a window and a count. Each has enough members for a narrow 95% interval, such as 9.1% to 13.2% for cart abandoners.

How do you test a tool in one hour?

  1. Minutes 0 to 15: build three segments from your real events: one "did", one "did not", one with a count and a window.
  2. Minutes 15 to 25: reconcile. Compare each previewed count with a count from a separate query on the same data. A difference above 1% to 2% needs an explanation (test users, bots, identity mode).
  3. Minutes 25 to 40: change the data. Trigger an event as a test user and time how long until membership changes. Compare this with the refresh mode in the criteria table.
  4. Minutes 40 to 50: reuse it. Apply a segment to a funnel and a retention table.
  5. Minutes 50 to 60: check exit. Export membership or receive a change notification, then delete a test person and confirm they leave.

What privacy checks apply to segments?

Building a segment from someone's behaviour is a form of profiling. GDPR Article 4(4) defines profiling as automated processing of personal data to evaluate personal aspects, including preferences, behaviour and location. Article 21(2) gives people the right to object at any time to processing, including profiling, for direct marketing, and Article 21(3) says that once they object their data must no longer be processed for that purpose. So three checks belong in any evaluation:

  • Can you honour an objection? A person who objects must leave every marketing segment.
  • Does the segment need personal data at all? Anonymous or aggregate segments avoid most of the problem.
  • Is the purpose written down? A segment built for product analysis should not quietly become a marketing list.

Common mistakes when choosing a segmentation tool

  • Choosing by attribute count. Fifty demographic fields are less useful than three well-defined behavioural rules.
  • Skipping "did not". Lapsed and abandoned segments need absence rules.
  • Trusting the preview. Reconcile it with a separate count.
  • Building tiny segments. Under a few hundred members the conversion rate is too noisy to act on.
  • Ignoring identity. If anonymous visitors cannot be members, a segment of "visitors who viewed pricing" is empty for logged-out traffic.
  • Counting overlapping segments as separate people.
  • Forgetting the exit. Confirm you can remove a person from every downstream list when they object, delete their account or stop qualifying, not only add them when they qualify.

Building event-based segments in TapCub

Take the five sample segments. You want them without exporting data to another system first.

  1. Identify people. Anonymous visitors have no profile and are never segment members, so call tapcub.login when someone signs in; they then appear in All-platform analytics › Users › People with a timeline.
  2. Build the segment. In All-platform analytics › Users › Segments & tags, create a rule segment from conditions such as did or did not do an event, a property, first seen, active, or being in another segment, preview the count, then save. Live segments add people at collection time (just did an event, visited a page, arrived from a channel); static segments are a pasted list of ids; RFM segments use ten buckets and record moves between them.
  3. Watch it. A rule segment refreshes daily or on demand, keeps a 30-day size trend and records who joined and who left.
  4. Reuse it. Analyze applies the segment as a cohort in analysis, so the same audience filters a funnel in All-platform analytics › Behavior › Behavior analysis.
  5. Know the boundary. If a webhook is configured, TapCub emits a membership change, but TapCub does not send email or buy ads. It defines and measures audiences; sending belongs to the system you connect.

FAQ

What are audience segmentation tools?

Software that defines groups of people by rules, keeps those groups current as behaviour changes, and lets you analyse them or pass them to another system. Behaviour-based ones build rules from events, such as did or did not do something in a time window.

What should you look for in an audience segmentation tool?

Check ten things: event rules with did and did-not, rolling windows, update mode, identity handling, size preview and trend, reuse in funnels and retention, export, privacy controls, limits and explainability. Weight them by your needs and score each option.

What is a behavioural segment?

A group defined by actions rather than attributes, such as users who added to cart in the last 7 days and did not buy. Each rule combines an event, a window and sometimes a count.

How big should a segment be?

Large enough that rates are stable. In the sample, 900 people at 11% gave an interval of 9.1% to 13.2%, which supports action; 40 people with 4 conversions gave 4.0% to 23.1%, which does not.

Does segmentation raise privacy issues?

Yes. Segments built from behaviour are profiling under GDPR Article 4(4), and for direct marketing people may object at any time (Article 21). Make sure objections remove people from segments and keep the purpose of each segment written down.

Sources

Was this article helpful?

We use your answers to fix and expand this page.

Sophia Bennett

Product Adoption Consultant

Explores customer onboarding, live conversations, and self-service, turning common usage questions into practical guidance and improvements.

Build segments from events in one project

Rule, live, static and RFM segments, a 30-day size trend, and one-click use in funnels and retention.

See your data clearly. Find your growth.

Every click, backed by data. Install one line of code and see your first numbers in a minute.

No credit card · Free plans for analytics and chat · Cookieless analytics