Install any skill in seconds. Free to start, no credit card required.
Get Started Free →When the user wants to identify, evaluate, or prioritize potential investors for a fundraising round. Also activates when the user asks "who should I pitch?", "find me investors", "build an investor list", or mentions VC/angel targeting.
.claude/skills/mkurman-investor-research/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-21 | ✗→✓ | ▲ Improved | 74% | 0% |
| case-19 | ✓→✗ | ▼ Worse | 13% | 0% |
| case-01 | ✓→✓ | = Same ✓ | 75% | 0% |
| case-02 | ✓→✓ | = Same ✓ | 32% | 0% |
| case-03 | ✓→✓ | = Same ✓ | 54% | 0% |
|---------|-------------|------------|------------|-------------|-----------|-----------|-------|
Followed by a "Conflicts" section listing excluded firms and why.
Followed by a "Research Gaps" section listing anything that could not be verified and needs the founder's input.
## Frameworks & Best Practices
### Investor Qualification Criteria (The 7-Point Filter)
1. **Stage fit** — Does the firm invest at the founder's current stage? A Series B fund will not lead a seed round. This is the first filter and it is binary: pass or fail.
2. **Sector focus** — Does the firm have a stated thesis or track record in the founder's sector? Look at their last 10 investments, not just their website copy.
3. **Check size match** — Does the firm's typical check size align with what the founder needs? A $2B fund rarely writes $500K checks. A $50M fund rarely leads $20M rounds.
4. **Portfolio conflicts** — Does the firm already have a company in the same space? This is the most common reason pitches are dead-on-arrival. Check every portfolio company, including quiet ones.
5. **Fund vintage** — Is the firm actively deploying from a recent fund? A fund raised 4+ years ago is likely in harvest mode and not writing new checks. Prefer firms that closed a fund within the last 18 months.
6. **Geographic relevance** — Some firms only invest locally. Others require board seats that demand proximity. Remote-friendly firms have expanded, but geography still matters for many funds.
7. **Partner-level interest** — Is there a specific partner whose background, interests, or public writing aligns with the startup? Pitching the right partner at the right firm matters as much as pitching the right firm.
### Tiering Framework
- **Tier 1**: Matches on 6-7 of the criteria above. The firm has invested in adjacent companies, the partner has spoken publicly about the space, and a warm intro path exists. Pursue first.
- **Tier 2**: Matches on 4-5 criteria. Good fit on stage and sector but may lack a warm path or have a slightly mismatched check size. Pursue in the second wave.
- **Tier 3**: Matches on 3 criteria. Acceptable as backfill if the round needs more participants. Do not spend significant time here until Tier 1 and 2 are exhausted.
### Sourcing Investor Information
- **Crunchbase / PitchBook**: Fund size, recent investments, portfolio companies.
- **Firm website**: Stated thesis, partner bios, blog posts that reveal focus areas.
- **Twitter/X and Substack**: Many partners publish their current interests publicly. Recent posts are a better signal than old "About" pages.
- **SEC filings**: Fund size from Form D filings when not publicly disclosed.
- **Portfolio founder back-channels**: The single best diligence on an investor is talking to founders they have backed — both successes and companies that struggled.
### Common Mistakes to Avoid
- **Spraying 200 cold emails** — Fundraising is a funnel. 30 well-targeted, well-introduced conversations beat 200 cold ones.
- **Ignoring portfolio conflicts** — Founders waste weeks pitching firms that will never invest because of a conflict.
- **Pitching the wrong partner** — At multi-partner firms, the wrong partner will say "interesting, let me introduce you to my colleague" at best, or just pass.
- **Targeting only brand-name firms** — Tier 2 and emerging funds are often faster to decide, more founder-friendly, and more willing to lead at earlier stages.
- **Not tracking your pipeline** — Use a simple spreadsheet or CRM: investor name, status (researching / intro requested / meeting scheduled / pitched / passed / term sheet), and next action.
### Angel Investor Considerations
- Angels decide faster (days, not weeks) but write smaller checks ($25K-$250K typically).
- Look for angels with operational experience in your sector — they add value beyond capital.
- Angel syndicates (AngelList, etc.) can aggregate small checks into a meaningful allocation.
- Be cautious about taking angel money from potential acquirers or competitors without understanding the signaling implications.
## Related Skills
- `pitch-deck` — tailor the deck narrative based on what specific investors care about
- `fundraising-email` — write targeted outreach once the investor list is built
## Examples
**Example prompt**: "We're raising a $2.5M seed round for a developer tools company based in SF. Help me build an investor list."
**Good output snippet** (one Tier 1 entry):
> | Boldstart Ventures | Ed Sim | Pre-seed/Seed | Developer tools, infrastructure | $1-3M | $160M Fund IV (2023) | None | Ed is active on Twitter re: dev tools; check if any portfolio founders overlap with your network | Led seed in [similar company]; blog post on "Why developer experience is the next platform shift" |
**Example prompt**: "I have a list of 15 VCs I want to pitch. Can you help me prioritize?"
**Good output approach**: Run each firm through the 7-point filter against the founder's startup context. Re-tier the list. Flag any portfolio conflicts the founder may have missed. Identify the 5 to pitch first and suggest the outreach sequence.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-06 | fail→fail | 14,395 | 9,964 | -31% | 1 | 1 | 0% | 1,965 | 2,560 | +30% | 0 | 0 | — |
case-01 | pass→pass | 11,557 | 12,257 | +6% | 1 | 1 | 0% | 1,663 | 2,917 | +75% | 0 | 0 | — |
case-02 | pass→pass | 19,965 | 18,964 | -5% | 1 | 1 | 0% | 3,171 | 4,178 | +32% | 0 | 0 | — |
case-03 | pass→pass | 16,851 | 17,302 | +3% | 1 | 1 | 0% | 2,502 | 3,854 | +54% | 0 | 0 | — |
case-04 | fail→fail | 16,809 | 14,029 | -17% | 1 | 1 | 0% | 2,425 | 3,278 | +35% | 0 | 0 | — |
case-05 | fail→fail | 17,152 | 15,804 | -8% | 1 | 1 | 0% | 2,594 | 3,594 | +39% | 0 | 0 | — |
case-07 | pass→pass | 16,339 | 15,668 | -4% | 1 | 1 | 0% | 2,484 | 3,404 | +37% | 0 | 0 | — |
case-08 | pass→pass | 15,550 | 13,085 | -16% | 1 | 1 | 0% | 2,277 | 3,137 | +38% | 0 | 0 | — |
case-09 | pass→pass | 11,637 | 6,089 | -48% | 1 | 1 | 0% | 1,772 | 2,170 | +22% | 0 | 0 | — |
case-10 | pass→pass | 16,248 | 14,722 | -9% | 1 | 1 | 0% | 2,360 | 3,409 | +44% | 0 | 0 | — |
case-11 | pass→pass | 15,522 | 11,806 | -24% | 1 | 1 | 0% | 2,147 | 3,035 | +41% | 0 | 0 | — |
case-12 | pass→pass | 16,454 | 15,149 | -8% | 1 | 1 | 0% | 2,415 | 3,492 | +45% | 0 | 0 | — |
case-13 | pass→pass | 11,624 | 10,125 | -13% | 1 | 1 | 0% | 1,692 | 2,595 | +53% | 0 | 0 | — |
case-14 | pass→pass | 15,127 | 20,780 | +37% | 1 | 1 | 0% | 2,247 | 4,302 | +91% | 0 | 0 | — |
case-15 | fail→fail | 15,329 | 16,709 | +9% | 1 | 1 | 0% | 2,463 | 3,792 | +54% | 0 | 0 | — |
case-16 | pass→pass | 14,777 | 17,122 | +16% | 1 | 1 | 0% | 2,099 | 3,657 | +74% | 0 | 0 | — |
case-17 | pass→pass | 14,338 | 13,533 | -6% | 1 | 1 | 0% | 2,148 | 3,272 | +52% | 0 | 0 | — |
case-18 | pass→pass | 10,229 | 4,183 | -59% | 1 | 1 | 0% | 1,445 | 1,838 | +27% | 0 | 0 | — |
case-19 | pass→fail | 15,518 | 8,172 | -47% | 1 | 1 | 0% | 2,208 | 2,488 | +13% | 0 | 0 | — |
case-20 | pass→pass | 17,649 | 16,064 | -9% | 1 | 1 | 0% | 2,601 | 3,640 | +40% | 0 | 0 | — |
case-21 | fail→pass | 14,397 | 16,798 | +17% | 1 | 1 | 0% | 2,113 | 3,667 | +74% | 0 | 0 | — |
case-22 | pass→pass | 16,655 | 16,210 | -3% | 1 | 1 | 0% | 2,398 | 3,627 | +51% | 0 | 0 | — |
case-23 | pass→pass | 15,488 | 16,158 | +4% | 1 | 1 | 0% | 2,399 | 3,695 | +54% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 23 cases were attempted. The headline lift of 0 percentage points is the difference between those two pass rates over the 23 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.