Talk notes
Repository creation is a quietly leading indicator that most analysts ignore because it looks like noise on a daily granularity. On a 90-day rolling window, the pattern emerges clearly. Companies that are about to scale engineering hire the repos before the engineers. They scaffold the infrastructure first — the new microservice template, the deployment-pipeline repo, the schema-registry repo — and then they hire to fill it.
The infrastructure scaffold has a telltale signature. A scaffolding repo will have an initial commit by a senior engineer, a clean directory structure (no half-finished branches), comprehensive README, and very few subsequent commits in the first 30 days. It looks like a parking spot waiting for the team. Compare that to a real product repo, which shows messy iteration, rapid commit cadence, and feature branches multiplying. The scaffold pattern, when it appears in clusters of 3–5 across an org in a 90-day window, almost always precedes a publicly-announced hiring round by 30–60 days.
The most common false positive is fork-vendoring — companies that mirror 100 OSS dependencies for compliance or vendoring reasons. We filter this by requiring a non-zero commit count from non-bot authors within the first 60 days of repo creation. That single filter removes 87 percent of fork-noise without sacrificing real product-expansion signal.
The strongest composite is repo expansion combined with sustained commit velocity. An org that's both shipping faster (commit velocity z-score above 2.0) and creating new repos (3× expansion versus baseline) is almost always either pre-fundraise scaling or post-fundraise execution. Either way, it's worth watching.