Segmentation

First PublishedMay 24, 2026Last UpdatedMay 25, 2026ByAtif Alam

The decision this page enables: which dimensions to segment your market on, and whether the segments you end up with are good enough to act on.

What segmentation is

Segmentation is the act of dividing a market into groups whose members behave similarly enough that you can address them with a coherent mix — and differently enough from other groups that the different treatment is worth the effort.

That last clause is what most teams skip. Slicing the market into ten groups that you then treat identically is not segmentation; it’s bookkeeping. A real segment changes at least one of the following: the message you use, the price you charge, the channel you reach them through, the feature emphasis you ship, or the success criteria you measure.

What “good” segments look like — the MSADA test

Before you commit to a set of segments, score each one against these five criteria. The mnemonic is MSADA:

Criterion	Question	Pass / fail
Measurable	Can we count members and track their behavior?	If you can’t size it, you can’t budget for it.
Substantial	Is the segment big enough to justify a distinct treatment?	Rule of thumb: ≥10% of your TAM, or ≥1% of an enormous TAM.
Accessible	Can we actually reach these people with a channel we can afford?	”C-suite at Fortune 100” is real but mostly inaccessible without enterprise sales.
Differentiable	Does the segment respond differently to mix changes?	If two segments react the same way to your price and message, they’re one segment.
Actionable	Can we design a distinct mix for them with our current resources?	A 12-segment strategy you can’t staff is worse than 2 segments you can.

If a candidate segment fails three or more criteria, it’s not a segment — it’s a description. Collapse it into another segment or drop it.

The flow from “raw market” through dimension choice and the MSADA filter to a working segment list looks like this:

flowchart TD
    M[Your market] --> D1[Demographic]
    M --> D2[Firmographic]
    M --> D3[Geographic]
    M --> D4[Behavioral]
    M --> D5[Psychographic]
    M --> D6[Needs-based / JTBD]
    D1 --> V{MSADA test<br/>Measurable · Substantial · Accessible<br/>Differentiable · Actionable}
    D2 --> V
    D3 --> V
    D4 --> V
    D5 --> V
    D6 --> V
    V -->|Pass| S[Working segment<br/>name it, size it, own it]
    V -->|Fail 3+| X[Discard or merge]
    S --> T[Targeting]

You don’t have to use all six dimensions — most working segmentations use two layered (one primary dimension + one overlay). The diagram is the menu; the next sections are how to pick.

The six dimensions you can segment on

Most segmentation in the wild uses one of these six lenses, alone or in combination:

Dimension	What it captures	Strongest fit	Page
Demographic	Age, gender, income, education, family stage, occupation	B2C	Demographic
Firmographic	Company size, industry, stage, geography, tech stack	B2B	Firmographic
Geographic	Country, region, climate, urban/rural	B2C and localized B2B (covered briefly below)
Behavioral	What they do — usage frequency, recency, buying journey stage, benefits sought	Both, and the most actionable lens for digital products	Behavioral
Psychographic	Values, attitudes, lifestyle, personality	B2C primarily; B2B buyer-mindset overlay	Psychographic
Needs-based (JTBD)	The job the customer is hiring you for	Both, increasingly the recommended primary lens	Needs-based

A note on geographic segmentation

Geographic segmentation gets a paragraph rather than its own page because, for most digital products, it’s a secondary overlay on top of one of the other dimensions — not a primary segmentation lens. Geography matters when language, payment infrastructure, regulation, or shipping cost varies meaningfully (e.g. EU GDPR exposure, LATAM payment methods, APAC working hours). For most early-stage SaaS, “EU vs US vs RoW” is enough granularity; you don’t need a country-level segment until you’re investing in country-level GTM.

B2B vs B2C cheatsheet — which dimensions to start with

Stage	B2B (SaaS / digital products)	B2C (apps / DTC)
Primary	Firmographic	Demographic
Layered on top	Needs-based (JTBD) + Behavioral	Psychographic + Behavioral
Useful overlay	Tech-stack and stage (subset of firmographic)	Geographic + occasion-based
Usually skipped first time	Demographic, Psychographic	Firmographic

Most teams stop at one dimension. The right answer is almost always two layered: a primary dimension that defines who they are, and a needs-based or behavioral overlay that captures what they’re trying to do. The combination is what produces segments that actually behave differently.

How to actually create segments — five methods

1. Manual bucketing. Pick 2–3 dimensions, define cut points, label the resulting cells. Crude but fast — fine for the first pass. Example: “10–49 person teams × already-using-Slack.” Most early-stage segmentation should start here.
2. RFM scoring (Recency, Frequency, Monetary) — see Behavioral. Best when you have transactional or usage data and need to prioritize who to invest in next.
3. K-means or clustering on multi-dimensional behavioral data. Useful when you have many variables and want the data to tell you the cluster boundaries instead of guessing. Risk: clusters can be statistically clean but commercially meaningless. Always validate against MSADA.
4. JTBD-driven — build segments around the job customers are hiring the product for. See Needs-based. Often the highest-signal approach when you have rich interview data.
5. Persona-derived — start from your Buyer Personas and ladder back up to segments by grouping personas that share buying behavior. Useful when the marketing team already lives in persona-land.

Combining dimensions — the segmentation grid

The most common production setup is a two-dimension grid — typically a primary dimension on one axis, and either behavior or job on the other. Example grid for a SaaS workspace:

                          │ Replacing Notion+Slack │ Starting from scratch │
─────────────────────────┼────────────────────────┼───────────────────────┤
 1-9 person teams        │ A: power-users         │ B: new founders       │
 10-49 person teams      │ C: budget refugees     │ D: scaling teams      │
 50+ person teams        │ E: enterprise PoC      │ F: out of scope       │

Cells become candidate segments. Score each against MSADA, drop the failures, name the survivors, and you have a working segmentation.

Template — MSADA segment scorecard

Use this for every candidate segment before you commit to it:

Segment name:        [short label, ideally 2-3 words]
One-line definition: [who they are, in plain language]
Estimated size:      [count or % of TAM]
Why this segment:    [hypothesis — what they share that matters]

MSADA score (1-5 each):
  Measurable:      __  - how we count them: [data source]
  Substantial:     __  - estimated size:    [number]
  Accessible:      __  - channel to reach:  [channel]
  Differentiable:  __  - what changes:      [mix element]
  Actionable:      __  - distinct mix:      [yes/no, why]

Total (out of 25): __

Verdict:  [keep / merge with X / drop]
Rationale: [one sentence]

A score below 15 usually means the segment isn’t real yet — or you’ve defined it on the wrong dimensions.

Metrics to track

• Number of active segments — most early-stage teams should have 2–3, not 8. More segments without more headcount is over-segmentation.
• Segment penetration rate — your customer count in segment ÷ estimated segment size. Floor: 1% penetration before you can claim “we serve this segment.”
• Cross-segment behavior delta — do retention, ARPU, CAC, or NPS actually differ across segments? If not, your segmentation is cosmetic. Target: at least a 25% delta on at least one of (retention, ARPU, win rate).
• Segment-message-fit score — for each segment, what % of inbound leads use language consistent with that segment’s stated pain? Healthy: ≥60%.
• Segment lifespan — how often you re-validate. Rule of thumb: full re-validation annually, lightweight check quarterly.

Worked example — SaaS workspace (B2B)

The team has finished customer-discovery interviews. They have 18 interview transcripts and CRM data on 240 trial signups. Here’s the first-pass segmentation:

Dimensions used: firmographic (company size + stage) × needs-based (the job they’re hiring the workspace for).

Candidate grid:

                                    │ "stop losing context"   │ "look organized to investors"  │
                                    │  job                    │  job                           │
────────────────────────────────────┼─────────────────────────┼────────────────────────────────┤
 5-15 person seed/Series-A startups │ S1: scaling product     │ S2: pre-Series-B prep          │
                                    │      teams              │                                │
 16-49 person Series-A/B startups   │ S3: ops-led teams       │ S4: out of scope (drop)        │
 50+ person startups                │ S5: enterprise PoC      │  -                             │

After MSADA scoring, S1 and S3 pass cleanly, S2 is interesting but small, S4 fails Actionable, S5 fails Substantial for the current stage. Working segments: S1 (5–15 person scaling product teams) and S3 (16–49 person ops-led teams), with S2 noted as a future-watch segment.

Worked example — Consumer fitness app (B2C)

Dimensions used: demographic (age + life stage) × needs-based (the job).

                              │ "feel less anxious   │ "lose weight by a │ "stay in shape   │
                              │  after work"         │  deadline"        │  while traveling"│
──────────────────────────────┼──────────────────────┼───────────────────┼──────────────────┤
 22-34, urban, no kids        │ F1                   │ F2                │ F3               │
 35-49, parents               │ F4                   │ F5                │ F6 (drop)        │
 50+                          │ -                    │ F7                │  -               │

After MSADA: F1 (job-anxiety reduction × young urban) and F5 (deadline-driven weight loss × parents) score highest. F3 looks attractive but fails Accessible at the current marketing budget (travel-fitness channels are expensive). Working segments: F1 and F5.

Note how the same product ends up with different segments in B2B vs B2C — and how neither primary dimension alone (firmographic or demographic) was sufficient. The needs overlay is what makes the segments behave differently.

Common pitfalls

• Segmenting on whatever data is easy. ZIP code is easy; it usually predicts nothing. Pick dimensions because they predict different behavior, not because the data is sitting in your CRM.
• Over-segmenting. Six segments × six campaigns × six landing pages = nothing executed well. Start with 2–3.
• Confusing personas with segments. A persona is one person you’re talking to; a segment is the group they belong to. You can have multiple personas inside one segment (champion, economic buyer, end user).
• Defining segments by features used. Feature use is downstream of need; if you segment on it, you’re locked into your current product roadmap.
• Not labeling who owns each segment. Every working segment needs a named DRI in marketing and one in sales. Unowned segments quietly become unserved.
• Letting segments drift unmaintained. Markets shift. A segment definition from two years ago is almost certainly wrong now.

See also

• Targeting — picking which of these segments you’ll actually pursue.
• Buyer Personas — the people-inside-segments lens.
• Jobs to be Done — the foundation for needs-based segmentation.
• Customer Discovery Interviews — the raw material most segmentations are built from.
• Strategy: ICP — the cross-functional artifact that summarizes “the segment we’ve targeted.”