The research stack
- LinkedInFind and qualify people with relevant experience.
- Gmail or OutlookCarry detailed invitations and follow-up.
- JustSendItConnects both communication channels to Claude.
- Separate Notion connectorStores participants, evidence, and next actions.
- ClaudePersonalises invitations and compares the findings.
Run a small research sprint
Keep evidence separate from interpretation.
- 1
Define
Write the decision, question, participant profile, and disconfirming evidence.
- 2
Find
Use LinkedIn to build a small relevant cohort.
- 3
Invite
Send an individual LinkedIn message or detailed email; say it is research, not a disguised pitch.
- 4
Record
Update status, response, evidence, and next action in Notion.
- 5
Synthesize
Compare themes, contradictions, and segment differences.
Notion research database
One row per participant.
- Person and company
- Why relevant
- Channel
- Status
- Last contact
- Next action
- Evidence
- Themes
- Opt-out
Use the same idea in two settings
Personal
Learn before a high-effort decision.
- Career change
- Community service
- Side project
Business
Preserve evidence behind product and market decisions.
- Problem interviews
- Win/loss research
- Category research
Limits
- The official Notion MCP uses interactive OAuth; treat this as session-based reconciliation, not guaranteed headless sync
- No automated multi-message campaign sequencing or LinkedIn scraping
- Keep consent, relevance, and opt-outs in the participant record
Copy these prompts
- “Find 20 people who match this research brief. Explain each fit in one evidence-based sentence and add them to Notion.”
- “Reconcile new LinkedIn and email replies with Notion, then show unmatched responses and contradictions in the evidence.”
Common questions
- Does JustSendIt connect directly to Notion?
- No. Connect the official Notion MCP separately in Claude and use it alongside JustSendIt.
- Is Notion synced in real time?
- Not by this setup alone. Update it during working or scheduled sessions and run a reconciliation step.