Skip to content
Investor Connect
Trust

How ranking works

The engine is deterministic — same inputs, same output, every time. No LLM in the ranking loop. No black box. Every match cites the specific signals that produced it.

The seven signals

35%
Sector overlap
Matches your sector words against every tag on the investor (curated + free-text). Higher confidence sources weigh more.
20%
Stage match
Exact stage match scores full; adjacent stages (pre-seed↔seed) score partial — reflects real-world investor behavior.
15%
Check size
Your target sits inside the investor's known check window? Full score. Outside? Decays smoothly.
10%
Geo
Exact state ≫ same country ≫ US-inferred. Investors without location get neutral weight.
10%
Portfolio overlap
Existing portfolio companies matching your sector + stage. Real evidence of thesis fit.
5%
Fund
Sanity check on check-to-fund ratio (0.1%–5% is a healthy seed check).
5%
Partner
Optional. If you name a partner, we boost investors who match by name or firm.

Weights sum to 100%. They're tuned deterministically and disclosed here — no hidden dials.

The two gates

Gate 1
Fit

Ranking must differentiate. A great match ranks materially higher than a poor one. We benchmarked ≥ 8/10 cohort queries pass this before releasing.

Gate 2
Why

Every match must carry a specific, evidence-grounded reason — not "generalist investor." Each why-string cites ≥ 2 signals.

Where the data comes from

Only public sources — regulatory filings, firm websites, and curated public directories. We do not use paid or proprietary data providers, and we do not scrape private information. Every enrichment field carries a source and a date so you can trace it.

Try it nowRead the privacy statement