Purchase intent data is a billion-dollar industry built on a simple truth: if you know a company is actively evaluating solutions in your category, you can reach them before your competitors do.
Most purchase intent signals are backward-looking — web traffic, content consumption, review site visits. By the time a company is reading G2 reviews, they're already mid-evaluation. The sales team that reaches them first wins, and that's rarely you.
GitHub is different. Engineering activity on GitHub reflects business decisions that haven't been announced yet. A startup that just closed a Series A is hiring engineers this week, not next quarter. New product bets show up as new repositories before they show up in press releases. Scaling teams generate GitHub signals 30 to 90 days before any public indication of growth.
Here are the five GitHub signals that most reliably predict B2B purchase intent — and how GitLeads scores each one.
Signal 1: Rapid Team Growth (New Org Members in 90 Days)
The single most reliable predictor of B2B purchase intent is headcount growth. When a startup goes from 6 to 14 engineers in 90 days, something happened: a funding round closed, a major customer signed, a product bet paid off. That inflection point is your window.
GitLeads tracks first-time contributors across every repo in an organization — contributors with zero prior commit history in that org. Across 50,000+ monitored GitHub organizations, new contributor additions in a 90-day window have the strongest correlation with companies entering active vendor evaluation.
Why it predicts buying: Growing teams need new tools. A team of 6 can get by on what they have. A team of 14 needs to standardize: communication, observability, CI/CD, security, data infrastructure. Engineering leaders at scaling companies are actively evaluating and buying in the first 90 days after a hiring surge.
Real-world pattern: A fintech startup's GitHub org shows 8 new contributors in 60 days — going from 5 to 13 engineers. No announcement, no LinkedIn posts yet. Three weeks later: a Series B press release. By then, their infrastructure vendors are already set. The sales teams that caught the GitHub signal reached them before the announcement. Everyone else is cold-emailing a company that just finished its buying cycle.
GitLeads score weight: 30%. Highest weight in the composite. New humans on payroll are the hardest signal to fake.
Signal 2: New Public Repos (Product Expansion)
Every new public repository is a declaration: we are building something new. New repos mean new products, new services, new technical bets. Companies in expansion mode are companies buying new tools.
GitLeads filters for repos with actual commit activity (not empty initialization repos) created in the last 60 days, weighted by contributor count. A new repo with 5 contributors already committing is a major product initiative. A new repo with 1 commit is noise.
Why it predicts buying: New product lines create new infrastructure requirements. A company spinning up a new mobile app needs a new testing framework. A company launching a data product needs new pipelines and observability. A company building an AI feature is evaluating LLM vendors right now. New repos are a proxy for new budget allocation.
Real-world pattern: A B2B SaaS company's org creates 3 new repos in 45 days: one named "mobile-app", one "analytics-v2", one "api-gateway". Each repo has 3-4 contributors already active. This company is expanding in three directions simultaneously. Sales teams selling mobile testing, analytics, or API tooling have a warm opening — the company is clearly mid-build on problems they're solving.
GitLeads score weight: 15%. Strong signal when combined with new contributor growth. Weak in isolation (solo founders create lots of repos).
Signal 3: Commit Velocity Spikes (Shipping Mode)
Commit velocity — commits per day over a rolling 30-day window — is a measure of engineering throughput. A spike in commit velocity means the team is in shipping mode: a launch deadline, a customer commitment, a product milestone. Companies in shipping mode are also companies making infrastructure decisions under time pressure.
GitLeads weights velocity spikes against a 6-month baseline for each org. A startup that averages 12 commits/day for 6 months and suddenly hits 40 commits/day is a more interesting signal than one that's consistently shipped 40 commits/day for two years. We're looking for acceleration, not absolute level.
Why it predicts buying: Velocity spikes often precede or coincide with product launches, which drive new tool adoption. Engineering teams under deadline pressure are simultaneously identifying gaps in their toolchain. "We can't launch on this timeline with our current CI/CD setup" is a buying trigger that happens in the middle of a velocity spike.
Real-world pattern: A dev tools company shows a 3× commit velocity increase over 6 weeks. Their Twitter/X is quiet. No blog posts. But the GitHub signal is clear: something is shipping. Sales teams who reach out with "I noticed your team has been shipping hard lately" get meetings. The insight feels specific, even though it came from a public signal.
GitLeads score weight: 25%. Second-highest weight. Velocity is a real-time measure of team momentum.
Signal 4: Issue and PR Volume Acceleration (Team Scaling)
Issues and pull requests are organizational signals, not just engineering ones. High issue volume means a growing user base filing bugs and feature requests. High PR volume means multiple engineers working in parallel, which means the team is large enough to have concurrent workstreams. Acceleration in both means a team that's scaling its process, not just its headcount.
GitLeads tracks PR open rate, review rate (comments per PR), and issue velocity separately, then combines them into a collaboration density score. A team with 4 engineers averaging 3 PRs/day per person is running a different operation than a team of 20 averaging 0.2 PRs/day per person.
Why it predicts buying: High collaboration density at scale requires tooling. Code review tools, project management, async communication, documentation — all of these become buying decisions when a team hits certain collaboration thresholds. When PR review cycles slow down because there are too many open PRs, the team starts evaluating review tooling. When issues pile up unresolved, project management tooling gets purchased.
Real-world pattern: A security startup's GitHub org shows a 4× increase in PRs opened over 8 weeks, with a growing ratio of external reviewers (contractors or new hires reviewing code). The company isn't in the press. But their GitHub collaboration pattern looks like a team that just doubled. They're about to hit the threshold where manual processes break — and when they do, they buy tools to fix them.
GitLeads score weight: 20%. Especially valuable for identifying companies approaching scale inflection points where tooling purchases become inevitable.
Signal 5: Dependency Adoption Patterns (Tech Stack Investment)
What a company adds to their package.json, requirements.txt, or go.mod is a direct window into their buying roadmap. New dependencies reveal technical decisions that haven't been announced: adopting a new cloud provider's SDK, integrating a new data framework, adding monitoring libraries for a tool they just bought.
GitLeads scans public repos for dependency changes in the last 30 days, cross-referencing against a taxonomy of B2B software categories: observability, security, payments, data infrastructure, communication, AI/ML, testing. We flag orgs that are actively adding dependencies in categories that suggest adjacent buying activity.
Why it predicts buying: Dependency adoption is a leading indicator of adjacent purchases. A company that just added the Datadog SDK is probably paying for Datadog — and probably evaluating Datadog's competitors or adjacent tools (log management, APM, infrastructure monitoring). A company adding OpenAI dependencies is in the middle of an AI tooling evaluation. You can see their stack choices before they make their final decisions.
Real-world pattern: An e-commerce startup adds Stripe's SDK, a fraud detection library, and a new analytics framework in the same two-week window. They're building a checkout flow from scratch — which means they're evaluating every piece of their payments and analytics stack right now. A competitor to any of those tools has a narrow window to get in front of them before the integrations are finished and the decision is locked.
GitLeads score weight: 10%. Highly specific when it fires — tells you not just that a company is growing, but what they're actively building and buying. Filtered to avoid false positives from test repos and one-off experiments.
How These Signals Work Together
Each signal is useful in isolation. Combined, they're a purchase intent radar.
The highest-confidence buying signal is the combination of new contributor growth plus commit velocity spike plus new repos. That pattern means: the team is growing, they're shipping fast, and they're expanding scope. Every B2B vendor category that startup touches is a potential buy right now.
The second-highest signal is new contributors plus PR volume acceleration. That's a team that just scaled headcount and is figuring out how to work at their new size. Process tooling purchases are imminent.
Dependency adoption as a standalone signal is useful for category-specific outreach: if you're selling observability tooling and a company just added your competitor's SDK, that's a warm contact, not a cold one.
Putting It Into Your Workflow
GitLeads surfaces these signals every week in a ranked Top 50 list. The workflow is straightforward:
- Monday morning: Read the Top 50. Scan the signal breakdown for each org — which of the 5 signals are driving their score?
- Filter for ICP fit: New contributors + new repos at a TypeScript shop tells a different story than the same signals at a Python ML company. Know what your buyer's stack looks like.
- Reach out with specificity: "I noticed your team has grown significantly and you've been shipping new repos this month" is not generic. It's an observation backed by a real signal. That specificity is what turns a cold email into a response.
- Time it right: The signal is freshest in the first 2 weeks. GitHub data compounds slowly — after 60 days, everyone has noticed the growth. The window is now.
The compounding advantage is real: teams that run this workflow consistently build a mental model of which GitHub signal combinations predict buyers for their specific product. That's institutional knowledge that compounds into a systematic edge over time.
Sign up for the free GitLeads weekly Top 50 email — every Monday, the 50 GitHub organizations showing the strongest purchase intent signals from the previous week, ranked and ready to work.