A GitHub star is the cheapest signal in software and one of the least honest. ReadSignal's 2025 research puts the ratio between a repository's stars and its paying customers at roughly 1000:1, against about 1:1 for closed source. Only 1 to 3 percent of the people who star a project turn into real buyers, and the trip from first star to signed customer runs two to six months. A maintainer can watch a repository climb past nine thousand stars and still struggle to name nine accounts that actually pay.

That gap is where the sales conversation begins. Open source has quietly become a distribution channel, and a pull request is its first real touchpoint — a developer reading the code closely enough to patch it, then deciding whether the project behind it deserves a line in a budget. Few teams staff that first conversation, which is how a warm signal goes cold before anybody opens a spreadsheet. Treating attention as revenue is the oldest mistake in this funnel. Reading the funnel stage by stage, and being honest about where it leaks, is the advantage.

The funnel, stage by stage

StageWhat the signal really meansWhere projects stall
Star or watchCuriosity and bookmarks; only 1–3% ever buyReporting stars to the board as pipeline
First pull requestA developer serious enough to read and patch the codeNo route from contributor toward a trial
Issue and discussionEarly qualification of a genuine problemThreads that never reach a product decision
Free trial or freemium80% prefer to try before any purchaseHiding the trial behind a mandatory sales call
Paid adoption48% of developers influence or approve the spendAssuming the champion can sign alone

The money behind this funnel explains why anybody bothers. The 2025 Commercial Open Source report from the Linux Foundation, COSSA and Serena, built on more than 800 venture-backed companies and twenty-five years of data, found that commercial open source firms reach a median IPO valuation of 1.3 billion dollars against 171 million for proprietary peers — roughly seven times — and a median acquisition price of 482 million against 34 million, about fourteen times. Funding reshapes the community around the code: independent contributors grow by an average of 27 percent, dependent projects multiply eight times, and package downloads seven times. Attention does compound. It simply does not compound into customers on the same schedule.

Here is the part the growth charts leave out. Developer influence is real but easily overstated. Stack Overflow's 2025 survey of more than 49,000 respondents across 177 countries found that 48 percent had endorsed or shaped a technology purchase in the previous year, yet only 19.8 percent of those involved a major new addition to the stack — most influence is incremental tuning, not a fresh budget line. And the buying journey is stubbornly hard to watch: just 15 to 20 percent of it unfolds inside the tools a single vendor controls. The rest happens in repositories, forums and chat rooms that no dashboard sees. Some 78 percent of developers discover new tools through GitHub and Stack Overflow, and 65 percent say a vendor's own open source work affects which vendor gets picked. A contributor fixing a bug is not a lead captured in a system. That contributor is a person forming an opinion that will surface somewhere no vendor can instrument.

The fair reading is that most of the funnel leaks, and that is acceptable. The 1000:1 ratio is not a failure of open source. It is the price of a channel where 80 percent of buyers insist on trying the software free before paying anything. A program that wants revenue has to treat the leak as design input. Make the trial the product rather than the teaser. Give the contributor a clean path from a merged patch to a paid deployment. Measure the slow two-to-six-month conversion as a pipeline with stages instead of a vanity chart. The projects that do this turn a star into a conversation, and a conversation into a qualification, without pretending the first number is the last one.

That discipline matters more when the software is a runtime shared by several hosts. An open catalog used across different super-apps and messengers cannot afford to treat contributions as free marketing, because every patch that lands in a shared runtime is a trust decision some host will later have to defend to its own users. Governance — who may merge, who may revoke, which commercial rules sit behind the free core — is what turns a contributor funnel into something a host can rely on. CrossMiniApp is one place where that runtime and that catalog meet, built so the first pull request and the first paid deployment sit on one governed path. The sales conversation starts early. The trust has to start earlier.

Sources