What makes a good market segmentation?

Groups that differ measurably in behaviour or economics — conversion, cycle length, deal size, retention — and that can be identified on a record before you engage. Tidy boundaries are not the objective.

What are the common ways to segment?

Size, industry, region, motion and use case. Size is the most common because it is easy to obtain; use case is frequently the most predictive and the hardest to capture.

How many segments should you have?

As few as behave differently. Each segment carries the cost of separate targeting, content and process, so a segment that does not warrant different treatment should be merged.

How do you validate a segmentation?

Compare outcome measures across segments. If conversion, cycle length and retention are similar across two segments, they are one segment with two names.

Segmentation

Dividing the market into groups that warrant different treatment — different motion, pricing, coverage or support model.

The test

A segmentation is useful when the groups behave differently and when membership can be determined before you engage. Both conditions are required — a segmentation that predicts behaviour but cannot be identified in advance is an analysis, not an operating model.

The validation, stated simply

Compare conversion rate, cycle length, deal size and retention across segments. Where they are similar, the boundary is describing your org chart rather than your market.

Common bases and what they are worth

Segmentation bases

Basis	Ease of capture	Predictive power
Employee count or revenue	High	Moderate; a proxy for complexity
Industry	High	Varies; strong where regulation differs
Region	High	Strong where language or law differs
Motion (self-serve / sales-led)	High	Strong; drives economics directly
Use case	Low	Often the strongest, and rarely captured

Source: Assessment stated here

Building one

Start from outcome differences in existing data rather than from a framework. Look for where conversion, cycle length and retention actually diverge.
Test candidate boundaries against those outcomes. Keep the ones that separate; discard the ones that feel right but do not.
Check identifiability — whether the segment can be assigned from data available at record creation. If not, it cannot drive routing or targeting.
Size each segment. A segment too small to warrant separate treatment is an observation, not an operating unit.
Write down the assignment rule and apply it in the systems, or the segmentation exists only in the deck.

Where it goes wrong

Segments drawn to match team structure rather than market behaviour, which guarantees they do not predict anything.
Too many segments, each too small to support the content and process it nominally requires.
Segments defined on data not present at the moment of routing, so the operational systems fall back to a default.
Set once and never revalidated, so the boundaries describe a market that has moved.

RELATED TERMS

GTM (Go-To-Market)

The coordinated plan for how a product reaches buyers: which segments, through which motions, with which pricing, message and coverage.

Territory Design

Dividing accounts among sellers so each has a patch with enough potential to support a quota, and so coverage is neither duplicated nor absent.

Enterprise Sales

The motion serving large organisations, characterised by buying committees, procurement, long cycles and high contract value.

Ideal Customer Profile (ICP)

The definition of the accounts your product serves best, grounded in the observed characteristics of customers who buy quickly, stay, and expand.

COMMON QUESTIONS

What makes a good market segmentation?: Groups that differ measurably in behaviour or economics — conversion, cycle length, deal size, retention — and that can be identified on a record before you engage. Tidy boundaries are not the objective.
What are the common ways to segment?: Size, industry, region, motion and use case. Size is the most common because it is easy to obtain; use case is frequently the most predictive and the hardest to capture.
How many segments should you have?: As few as behave differently. Each segment carries the cost of separate targeting, content and process, so a segment that does not warrant different treatment should be merged.
How do you validate a segmentation?: Compare outcome measures across segments. If conversion, cycle length and retention are similar across two segments, they are one segment with two names.