Market Map

Maps a sector so the next four steps are cheap and auditable. The output is a decision document that tells the user where not to look.

Read first:

• references/market-map-method.md — the six sections and how each is built
• references/sourcing-sources.md — source types, for section 5
• references/research-ethics.md — terms of use, personal data, the lines

This skill produces no company list. Names you stumble across go in a scratch list at the end, and company-discovery enumerates properly once the map is approved.

Step 0 — Load the mandate

Read mandate-profile.md from the user's workspace.

If it does not exist: copy templates/mandate-profile.md into the workspace and walk the user through sections 2 to 7 at minimum before mapping. A map built without a mandate maps somebody else's sector.

🚧 HARD GATE — intermediary check

Check section 9 of the mandate profile:

• Principal buyer: proceed
• Fee-based intermediary: STOP. Tell the user plainly that this can trigger broker-dealer registration questions in the US and that counsel should clear the model first. Proceed only after they confirm it is cleared. Per references/research-ethics.md, this is not legal advice — it is the first question counsel will ask
• TBD or blank: stop and ask

Step 1 — Confirm scope

Restate back, from the mandate:

• The sector to map, and the geography
• The size box, so section 2 can be judged against it
• Deal type and ownership types in scope
• Known exclusions

Ask one question if the sector is broad: map the whole sector, or one segment of it? Mapping "industrial services" is three maps. Mapping "commercial fire and life safety in the Southeast" is one.

Wait for confirmation.

Step 2 — Read, do not search-and-summarize

Work in roughly this order, and cite everything:

1. Association publications, member magazines, annual reports
2. Trade press: consolidation coverage, state-of-the-industry pieces
3. Sponsor portfolio pages and add-on press releases
4. The largest players' own sites — segment language, service mix, footprint
5. Public company filings for any listed comparable; the risk factors section is unusually honest about sector dynamics
6. Licensing and certification registries, for counts

Rules while reading:

• Never write a sector fact from memory. If it is not on a page you read in this session, it is Not found
• Count things where counting is possible. A licence roster count with a stated method beats "highly fragmented" every time
• Label every estimate with its method in the same sentence
• Name sponsors and cite the portfolio page. "The sector is consolidating" with no names is not a finding
• Segment by how businesses earn, not by industry classification

Check in after the reading pass, before writing, with what you found and what you could not establish.

Step 3 — Write the map

Use templates/market-map.md. All six sections:

1. Segments — three to six, operating-based, each with a one-line profile and an in-mandate flag
2. Size distribution — counts by band, method stated, and an explicit answer to whether the mandate size box intersects where the mass sits
3. Fragmentation and consolidation — largest players and their ownership; named sponsors with platforms here; the add-on profile they buy; an estimated sponsor-owned share
4. The typical owner — generation, background, and two or three sourced public quotes. Professional sources only, no generalizations about age or health
5. Discovery surface — actual sources with URLs, ranked for this sector, with a terms-of-use note per source
6. Where not to look — segments, size bands, sponsor clusters, franchise networks, geographies

Plus the summary at the top with a one-line verdict: Source this sector, Source a segment only, or Do not source.

🚧 HARD GATE — before writing the file

Show the user:

• The summary and the verdict
• The segment table
• The sponsor list
• The three best sourcing angles
• The "where not to look" section in full
• Everything you could not establish, as open questions

Get approval. Then write.

Step 4 — Feed the map forward

Two hand-offs, both explicit:

1. Sponsor list → mandate profile section 8. Offer to add the named sponsors to the exclusions section. mandate-fit-scoring screens against it
2. Segments → mandate profile section 4. If the map revealed better segment language than the mandate had, offer to update it

Ask before editing the mandate profile. It is the user's file.

Step 5 — Report

Market map complete — Commercial fire & life safety, Southeast US

Verdict: Source a segment only

Segments identified:        4  (2 in mandate)
Size bands counted:         4  (method: state licence roster + association list)
Est. population in mandate: ~310 companies
Sponsors named:             6  (11 platforms, 40+ add-ons since 2022)
Est. sponsor-owned share:   ~25% of the in-mandate population
Owner quotes sourced:       3
Discovery sources found:    9  (2 prohibit automated collection — manual only)

Best three angles:
  1. State fire protection licence rosters (mandatory, dated, complete)
  2. NFSA/AFSA chapter member directories
  3. Trade show exhibitor lists, 2022-2026

Where not to look:
  - Residential-only operators (out of mandate)
  - Sub-$5M revenue band (~60% of population, below the size floor)
  - The 11 known sponsor platforms and their announced add-ons

Could not establish:
  - Margin norms in the inspection-recurring segment (no public source found)
  - Whether the two largest independents are family or partner-owned

Next: run company-discovery against angles 1-3.

Report what you could not establish. A map with no open questions has almost certainly filled gaps with plausible sector generalities.

Notes

• A map is not a list. If you find yourself writing company names into the map, stop. Note them in a scratch list and move on
• The test of a good map is what it excludes. If every company in the sector sounds like a target, the map has not done its job
• Three to six hours of reading is normal. A map produced in ten minutes is a summary of what the model already believed about the sector, which is exactly the failure mode this step exists to prevent
• Re-read the consolidation section before every new sourcing cycle. It goes stale fastest
• One map per sector per geography. Do not merge two sectors into one map to save time; the segment boundaries stop meaning anything