Segmentation (needs-based, from the corpus)

Split the market or user base into groups that differ in what they need, not just in what they look like. Real segments earn their keep by pointing at different product bets; demographic buckets that all want the same thing don't. Every segment here is defined by evidence of a distinct need and carries the customers' own voice.

Step 0 — Relevance & availability

Confirm this is segmentation work and Evermuse is connected (see using-evermuse Step 0). If disconnected, produce a framework-only model labeled ⚠ ungrounded and tell the user to authorize the MCP.

Two altitudes — pick one

• User segmentation (existing users, from feedback) → segments come from the corpus; save nature=evidence.
• Market segments (the broader addressable market, incl. non-users) → market-level cuts lean on context searches (industry/analyst signals) validated by customer evidence; the customer-derived parts still save as evidence. State which you're doing up front. Default to user segmentation unless the user is clearly exploring a market they don't yet serve.

Evermuse Grounding (required)

Follow ${CLAUDE_PLUGIN_ROOT}/skills/using-evermuse/SKILL.md.

Required — one parallel batch of searches. Verify the product (Rule 1), then fire in a single parallel batch:

• 3–4 evidence searches, each worded from a different angle (limit up to 50). Evidence comes back rich and varied — expect large, useful result sets.
• one guidance search and one context search (limit up to 50). These are usually sparse or empty; run them anyway and note when they're thin.

Read each response's digest — it reports how many more results exist. Use judgment on whether a query is worth pulling deeper (raise limit toward the 100 max and/or page with next_offset to avoid repeats), weighing payload size, remaining context, task complexity, and the value of the data. For deep pulls, consider spawning sub-agents — instruct them to return every citation with the same metadata the tools return (url, who_said_it, meeting_name, created_at) so you can still cite.

Optional — considered use. Once grounded, reach for the other tools only when they add value: a quote-angled search for verbatim voice, read_source for a single deep dive, inline citations (references/citations.md), and add_source to save the deliverable. Sources, citations, and saving are optional — not required.

Instructions

Segment $ARGUMENTS into 3-5 distinct groups.

1. Choose segmentation dimensions from the evidence — behavioral (how they use it), needs-based (the job), workflow context. Reject a demographic dimension unless the evidence shows it maps to a real difference in need.
2. Define segments so they're measurable, non-overlapping, and each points at a different product decision.
3. Characterize each segment with the structure below, quoting real customers. Where the source method says "if the user provides data" or reaches for market studies, use evidence searches (and context only for market-size color).
4. Validate distinctness — if two segments want the same things, merge them. Fewer real segments beats more fake ones.

Segment profile (each of 3-5)

Name & one-line characterization — memorable, and says what makes it different.

Size / weight — rough share of the corpus (accounts/mentions), labeled as a corpus estimate, not a market number, unless you have context data.

Jobs-to-be-done — the primary job, its context, frequency, and success criteria for this group.

Key pains & unmet needs — badged with the accounts they came from.

Representative voice — 1-2 verbatim quotes:

> "[verbatim quote]" — [Name], [Meeting], [Date] [`1`](URL)

Product fit — how well the product serves this segment today; the sharpest gap.

Differentiated value / positioning — what would move this segment most, and the messaging that resonates.

Priority — invest / maintain / de-prioritize, with a one-line rationale (growth, revenue, strategic fit vs. effort).

Close with a note flagging any segment that's underrepresented in the corpus. This pairs naturally with /evermuse:user-personas (a persona per segment) and market-sizing.

Further reading

• Merges the market-segments and user-segmentation methods from phuryn/pm-skills (MIT); grounding is Evermuse-native. See Crossing the Chasm for PMs.