GitHub Engineering Velocity Data, Methodology and Benchmarks

Data Methodology

GitDealFlow scans public GitHub organizations for 350+ venture-backed startups across 15 sectors.

Three core metrics: commit velocity (weekly commits), contributor growth (new contributors per month), and repo expansion (new public repos and code size growth).

Data refreshes every Monday based on the prior week's GitHub activity.

All data is publicly available and free via JSON API, CSV export, and MCP server.

Sector Benchmarks

AI/ML: 2.3x average commit velocity vs. all sectors. Fastest shipping sector.

Fintech: 1.7x average. Led by stablecoin and embedded lending infrastructure.

Climate tech: Fastest acceleration. Q1-Q2 2026 velocity grew 3x.

DevTools: Strongest contributor retention. Teams stay together longer.

Biotech: Slowest average commits but highest correlation with fundraise events.

Why This Page Exists

GitDealFlow is a public deal flow signal dataset: 350+ startup GitHub organizations across 15 sectors, refreshed weekly, with breakout teams surfacing 21 to 47 days before their round is announced. This page makes one part of that system legible: what it measures, how it is computed, and how to use it in a live sourcing workflow. The method is published end to end and falsifiable by design, with the working paper on SSRN and the dataset downloadable under CC BY 4.0.

Start Here

• Code-Side Sourcing: the category and its method
• Free momentum checker
• Research and datasets
• Pricing

A practical read-through of GitHub Engineering Velocity Data, Methodology and Benchmarks: the dataset behind this page refreshes weekly across 350+ organizations and 15 sectors, and every figure shown traces to a public GitHub REST API pull. That matters for two reasons. Reproducibility: any number here can be re-derived from primary sources, which is the standard the published methodology sets for itself. Timeliness: engineering acceleration precedes announcements, so this page follows the data cadence rather than the news cycle, and the freshness endpoint always reports the exact pull date.

If GitHub Engineering Velocity Data, Methodology and Benchmarks is your entry point, the fastest next steps are fixed: skim the glossary for the three or four terms that anchor the topic, open the research dataset to see the raw weekly snapshots behind the summary numbers, and run one live query against the free momentum checker with a company you already know well. Seeing the signal fire on a familiar name is the quickest way to judge whether code-side sourcing belongs in your own workflow.

One caveat worth stating plainly on GitHub Engineering Velocity Data, Methodology and Benchmarks: momentum is a leading indicator, not a verdict. A repository can accelerate for reasons that never become a fundraise, and a quiet quarter does not mean a team is failing. The disciplined use of this page is as one input in a stack, a way to rank where scarce diligence time goes, and a way to notice change early. The methodology page documents every limitation, including the bot filter, the two-period confirmation rule, and the sectors where coverage is thinnest.

From raw pulls to a deal-flow signal

Each metric earns its place by surviving a false-positive filter. Commit velocity is a percentage change across 14-day windows and only counts after two consecutive confirming periods, which removes release spikes and rebuild weeks. Contributor concentration is a Gini coefficient: it distinguishes a team scaling its contributor base from a project one resignation away from dead. Repository expansion counts new public repositories and code-size growth, which correlates with team growth before headcount is announced anywhere.

The composite is deliberately transparent: the Momentum Score is round(0.40·traction + 0.35·recency + 0.25·velocity), computed from public GitHub REST API data only. Breakout teams that clear the confirmation bar surface 21 to 47 days before their round is announced, and everything needed to check that claim, the dataset (CC BY 4.0), the JSON and CSV endpoints, and the OpenAPI spec, is public and free.