gtmskills.com

Reference file

Attendee research

attendee-research.md

Attendee research: query fan-out and profile format

How to research one attendee. Run one of these per attendee, in parallel across attendees. Before starting, pass in any known facts from the person's running file so the research reports only what's new.

Searches — Group 1 (run simultaneously)

  1. "[Name] [Company]" — broad web search, ~10 results. The primary query: role, title, background, company association.
  2. People search for "[Name] [Company]" — if your research tool can search the people indices of social platforms directly, use it; it is the most reliable way to find someone's LinkedIn profile and social presence.
  3. "[Name]" restricted to linkedin.com — the fallback for query 2. If both return results, prefer query 2 (richer people-index data).
  4. "[Name] [Company] interview OR podcast OR talk OR keynote" — where they share their actual thinking.

Email-enhanced variant: if you have the attendee's email (from a calendar invite), replace query 3 with "[first] [last] [email-domain]" restricted to linkedin.com. The email domain confirms the company and disambiguates common names far better than name alone.

Retry rule: fewer than 3 total results from Group 1 → drop any date filters and try name variations: "[First] [Last]", "[Full Name] [Company]", "[Full Name] [Title if known]".

Searches — Group 2 (after Group 1 returns)

  1. "[Name]" restricted to x.com, past month — recent opinions, announcements, what they care about right now.
  2. "[Name] [Company]" restricted to github.com, medium.com, substack.com — technical and thought-leadership content they authored.
  3. "[Name] conference OR speaker OR panel OR published" — appearances and published work.

Keep scope tight: roughly 7 searches and at most 2 page extractions per attendee. Running in two groups also keeps you under rate limits when several attendee threads run at once.

Extraction (max 2 per attendee)

  • If a LinkedIn profile URL surfaced, extract the page (render it if your tool supports it). Profile photos are not extractable — don't try.
  • If a recent interview or talk surfaced, extract the top result.

Name disambiguation

If the primary query returns multiple different people with the same name, filter by the company. Still ambiguous → stop and present the top candidates with their titles and companies for the user to pick. Never silently research the wrong person.

Cross-attendee overlap flags

When a thread discovers a workplace, school, board, or community for its attendee, check it against the other attendees' findings — "my attendee worked at that company 2019–2022, did yours overlap?" These overlaps feed the briefing's Relationship Map; hunting for them actively beats comparing results after the fact.

Return format (exact)

PROFILE:
Name: [Full name]
Title: [Current title]
Company: [Company name]
LinkedIn: [Profile URL if found]
Time in role: [Duration if found]
Location: [If found]

CAREER:
- [Previous role] at [Company] ([years])

EDUCATION:
- [Degree, Institution] (if found)

RECENT ACTIVITY:
- [What they've posted, shared, or spoken about — with dates]
- [Direct quotes when available — these are gold for conversation hooks]

INTERESTS & OPINIONS:
- [Topics they care about based on posts/talks/articles]
- [Positions they've taken publicly]

CONNECTIONS:
- [Notable past employers — flag any overlap with other attendees]
- [Organizations, boards, or communities they're part of]

SOURCES:
- [URL] — [what it contained]

Anything already in the known facts stays out of RECENT ACTIVITY — the profile carries the full picture; the research reports the delta.