You will evaluate new signals against an explicit thesis, classify their materiality, and document whether your recommendation holds — using a repeatable claim-check workflow instead of gut feel.
The hardest moment in analysis is not finding new information. It is deciding whether the new information matters to a recommendation you already socialized. Teams either over-react to every headline or under-react until a client forwards a competitor press release.
Prerequisites
- A written thesis: recommendation + 2–3 supporting claims + assumptions
- A list of thesis breakers you defined when you shipped the last brief
- New evidence captured in your source library with date and source type
Steps
1. Restate the current thesis in three lines
Write:
- Recommendation:
- Primary supporting claims: (numbered)
- Assumptions we are not testing this week:
If any of these exist only in someone's head, stop and write them down first.
2. Classify the new evidence
For each new signal, tag:
| Tag | Meaning |
|---|---|
| Fact | Verifiable event (launch, price change, filing) |
| Interpretation | Analyst or press inference |
| Rumor | Unconfirmed; treat as hypothesis |
Only facts can directly break a thesis. Interpretations trigger a "verify" task, not an automatic pivot.
3. Map evidence to claims
Ask: Which supporting claim does this touch?
- If none → immaterial for now (log it, do not rewrite the brief)
- If one → run claim-check (Step 4)
- If multiple → material — schedule a thesis review
4. Run claim-check against prior sources
For the affected claim:
- What did we believe before? (cite prior passage)
- What does the new source say? (cite new passage)
- Do they conflict, extend, or confirm?
Conflict → downgrade confidence on that claim; assess if recommendation still holds.
Extend → narrows uncertainty; update evidence bullet, not necessarily the recommendation.
Confirm → strengthens; note in weekly watch, no brief rewrite needed.
5. Score materiality
Use a simple gate:
- High: Would change the client's decision or our recommended action
- Medium: Changes a supporting claim but not the decision
- Low: Interesting context, no thesis impact
Only high materiality triggers a revised recommendation before the next scheduled review.
6. Document the thesis delta
One paragraph, every time you run this workflow:
Thesis delta ([date]): [Holds / Weakens / Strengthens / Reversed]. New evidence: [one line]. Affected claim: [#]. Action: [none / update brief / call client].
This log becomes invaluable across long engagements — partners see judgment over time, not just the latest slide deck.
Verification
The workflow works when:
- You can point to a thesis breaker that fired (or did not) when news broke
- Materiality calls are documented, not debated from memory in Slack
- Recommendations change before the client asks "does this still hold?"
Troubleshooting
Everything feels material. Re-read the decision statement — most news is context, not thesis-breaking.
Never update the thesis. You may have vague breakers; rewrite them as testable conditions.
Team disagrees on materiality. Resolve with the decision lens: would we change the recommended action?