Talk notes
Every quantitative product owes its buyers a precision number and a false-positive post-mortem. Ours: 62 percent precision at the 90-day horizon. That means of every 100 organizations the signal flags, 62 close a publicly-announced fundraise within 90 days, and 38 don't. The honest question is — what happens to those 38?
We did the post-mortem on the most recent 100 false positives in our panel. Of the 38 that didn't raise on the public timeline, 27 raised silently. That means an extension round, a secondary, a strategic check-in, or an SAFE that never made it to Crunchbase. Crunchbase coverage is increasingly leaky — the gap between what's actually happening in venture and what's publicly indexed has widened sharply since 2024. So our 'false positive' is often Crunchbase's blind spot.
Six of the 38 had a major product launch in the same 90-day window. That's a different kind of true positive — the engineering acceleration was real and predicted a real event, just not a fundraise. For some buyers (corp-dev, partnerships, sales), product launches are more actionable than fundraises anyway. The signal still has economic value for them.
Four of the 38 had no material event we could detect within 90 days. That's the genuine miss. Four percent of our flags are noise. We name it explicitly because we'd rather a buyer trust the bounded scope than discover the noise floor by accident. The buyer who sees a 4 percent noise rate and decides they can live with it is the buyer who's read this slide.