Skip to main content
Blog
Back to Blog

How to market an open source project in 2026

You made a repository public and almost nobody uses it. Most advice on promoting an open source project is a list of places to post and tricks for more GitHub stars, written a few years ago. Stars have since become easy to fake: a study presented at ICSE 2026 estimated six million suspected fake stars on GitHub and found that bought popularity fades within about two months. This guide covers what to market instead of stars, which audience to start with, a repository checklist using GitHub's own limits, which channel fits which kind of project, the promotion that backfires, a 30-day plan, a weekly launch ledger you can copy and how users turn into customers if the project is part of a startup. It is useful without a Soloop account.

The short answer

To market an open source project, write for one kind of user with one problem, make the README and quickstart get that person to a working result in minutes, and launch in one place where those users already talk. Then talk to the people who try it, fix what stopped them and launch again somewhere else.

Measure people, not stars. GitHub forbids automated starring and paid engagement, and research presented at ICSE 2026 found that fake stars give a project a short boost followed by lasting damage. Track who visits, who clones or installs, who opens an issue and who comes back, and keep the 14 days of traffic GitHub shows you before it disappears.

Why stars are a weak goal in 2026

A star is a bookmark. It costs one click, says nothing about whether the person used the project and can be bought. A research team led by Hao He estimated about six million suspected fake stars between 2019 and 2024, with a sharp rise in 2024. Most promoted short-lived malware repositories; the rest mainly promoted AI and LLM, blockchain, tool and tutorial projects. The promotion effect lasted under about two months, after which the fake stars became a liability.

Stars still matter as a public signal. Investors and lists watch them: Runa Capital's ROSS Index ranks open source startups by star growth, and in the third quarter of 2026 almost every project on it was an AI or agent tool. Treat stars as a lagging result of people finding the project useful, never as the target you optimize.

Fake stars

About six million suspected fake stars from 2019 to 2024, surging in 2024; most promote short-lived malware repositories, the rest mostly AI/LLM, blockchain, tool and tutorial repositories; the promotion effect lasts under about two months and then becomes a liability.

He et al., Six Million (Suspected) Fake Stars in GitHub (ICSE 2026)
GitHub's rules

Prohibits inauthentic interactions, rank abuse such as automated starring or following, engagement incentivized by rewards such as airdrops and secondary markets for inauthentic activity.

GitHub Acceptable Use Policies, section 4
Star growth as a market signal

The Q3 2026 ROSS Index ranks open source startups by GitHub star growth; nearly all 20 entries are AI or agent projects.

Runa Capital: ROSS Index Q3 2026

Decide who you are marketing to first

An open source project has three possible audiences, and they need different things. GitHub's maintainer panel put it simply: grow users first, because visible use is what attracts contributors. If the project belongs to a company, buyers come later still, from the users who depend on it at work.

Users, contributors and buyers need different things
DimensionWhat they need from youFirst move
UsersA plain statement of the problem it solves, a quickstart that works and proof that it is maintained.Rewrite the first screen of the README for one user and one problem, with a 10–20 second demo.
ContributorsContribution guidelines, a roadmap, labeled starter issues and maintainers who reply kindly.Add CONTRIBUTING, a short roadmap and a few well-described issues, then thank every first report.
Buyers (if you sell something)A reason to pay that does not break the free project: hosting, support, team features or compliance.Ask the people who use it at work what they would not want to run themselves.
Users and contributors

Name a specific problem and benefit; target a niche; share 10–20 second demos; keep the README about the user's problem; contributors follow organic user adoption and need guidelines, a roadmap and responsive, kind maintainers; measure long-term impact rather than downloads or stars.

GitHub Blog: Marketing for maintainers (Jul 2022)

Repository checklist before you promote anything

Every channel sends people to the same page. If the repository does not explain itself in the first screen, promotion only makes more people leave. Check each item first.

  • Description and first README line: who it is for and what it does, in plain words, without implementation details.
  • A demo near the top: a GIF or short video of 10–20 seconds showing the first useful result.
  • Quickstart: copy-paste install and one command or snippet that works on a clean machine. Test it from scratch before launch.
  • Topics: up to 20, lowercase letters, numbers and hyphens, 50 characters each. Use the words people search for, such as the language, framework and problem.
  • Social preview image: at least 640×320, ideally 1280×640, PNG, JPG or GIF under 1 MB, so links shared on X, Slack or Reddit look intentional.
  • A license file, so companies know whether they may use it.
  • CONTRIBUTING, a short roadmap and a few issues labeled for newcomers with enough context to start.
  • Signs of life: a recent release, answered issues and a changelog. Visitors check whether anyone is home.
  • Who uses it: real users or companies, named only with their permission.
  • A link to docs or a website if one exists, and how to get help.

Choose a launch channel by what the project is

Pick the one place where people with the problem already gather, show the project solving it and stay to answer questions. A second channel can wait until the first has taught you what people misunderstand.

Where the first users of an open source project usually are
DimensionWhere the first users areFirst move
Library or SDKThe framework's community, its issue tracker and discussions, and Stack Overflow questions your library answers.Answer existing questions where the library genuinely helps, and publish a short example that solves a common task.
CLI or developer toolHacker News, developer subreddits and X accounts that share tools.A Show HN people can try without signing up, with you in the comments for the first hours.
Self-hosted alternative to a paid productPeople complaining about the paid product's price, limits or privacy, and self-hosting communities.A clear comparison of what it replaces, what it does not do yet and how to run it.
AI agent tool, MCP server or pluginBuilders on X and GitHub, agent framework communities and AI assistant directories.A demo of one task the agent could not do before, and a listing in the relevant directory if one exists.
Open-core startupTeams already using the project at work.Talk to the people behind issues and installs from company domains before announcing a paid tier.

Promotion that backfires

These tactics either break a platform's rules or produce numbers that mislead you. Check your plan against the list before launch week.

  • Buying stars, trading stars with other maintainers or using tools that star automatically.
  • Giving rewards, airdrops or giveaways in exchange for stars, follows or other engagement.
  • Asking friends or colleagues to upvote or comment on your Show HN.
  • Posting a landing page, waitlist or sign-up form as a Show HN instead of something people can try.
  • Advertising in other projects' issues, or posting the same promotion across many repositories.
  • Scraping GitHub profiles to send unsolicited emails about your project.
  • Posting the same link in many subreddits or groups without reading their self-promotion rules.
GitHub

Section 4 bans inauthentic activity, rank abuse such as automated starring, reward-incentivized engagement and coordinated inauthentic activity; section 7 bans using GitHub information for spam such as unsolicited email; section 10 bans monetized or excessive bulk content in issues.

GitHub Acceptable Use Policies
Show HN

For something people can play with; not blog posts, sign-up pages, newsletters or lists; no landing pages or fundraisers; don't ask friends to upvote or comment; make it easy to try without signups.

Hacker News: Show HN guidelines

A 30-day launch plan

  1. Days 1–3: write one sentence naming the user, the problem and the result. Rewrite the repository description, first README paragraph and demo from it.
  2. Days 4–7: work through the repository checklist above. Ask two people who have never seen the project to follow the quickstart while you watch, and fix every place they get stuck.
  3. Days 8–14: find 20 existing conversations where someone has the problem, such as issues, forum threads or posts. Reply only where the project actually helps, and say that you made it.
  4. Days 15–21: make one public launch in the channel where those conversations happened, such as a Show HN or a demo post. Stay in the comments and write down every question and objection.
  5. Days 22–30: fix the most common objection, publish a release that says what changed, thank the people who reported problems and compare this window's ledger with the first one.

Copy the weekly launch ledger

GitHub shows visitors, clones, top referrers and popular content for the last 14 days only, and only to people with push access. The REST API returns the same 14 days, so copy the numbers every week or lose them. Package registries keep longer history: npm's download API answers for periods such as last week or a date range within 18 months, and PyPI publishes downloads to a public BigQuery dataset.

Downloads count installations, including automated ones, not people. Read them as a trend next to the human signals in the ledger: issues from strangers, first contributions and conversations with people using the project for real work.

Download Markdown
npm downloads

Point and range endpoints for last-day, last-week, last-month, last-year or a date range; queries limited to 18 months of data.

npm registry: download counts
PyPI downloads

Download events are streamed into the public BigQuery table bigquery-public-data.pypi.file_downloads.

PyPI Docs: BigQuery datasets

Read the numbers together

Illustrative example, not a benchmark: in the 14 days around a Show HN, a repository gets 2,400 unique visitors and 180 unique cloners, so 7.5% of visitors cloned it (180 ÷ 2,400). It also gains 150 stars, about 6 per 100 visitors, and 12 issues from people outside the team. After the README is rewritten around the most common question, the next 14 days bring fewer visitors, 2,000, but 200 unique cloners, or 10% (200 ÷ 2,000). Fewer people arrived, yet more of them tried it: the rewrite worked, even though stars grew more slowly.

Read stars in that context. A jump in stars with no change in cloners, installs or issues usually means attention without use. Growing issues from strangers usually means real use, even when they are bug reports.

From users to customers

If the project is part of a startup, the first customers are usually already in your issue tracker. Look for people using it at work, ask what they would rather not operate themselves and test one paid offer with them, such as hosting, support, team features or compliance help, before building a pricing page.

Keep the free project genuinely useful. Moving the quickstart behind a sign-up or removing features users rely on breaks the trust that brought them, and on Hacker News a sign-up wall disqualifies a Show HN.

Where Soloop fits

Soloop does not star repositories, publish packages or manage your GitHub project. It helps with the outreach around it, and you can start from a repository. According to the Soloop manual, it can research where people discuss a problem and return sources, post on X after you connect it with OAuth, prepare Reddit posts for you to publish yourself, send cold email through a connected Gmail account and prepare Google Search Ads for your review and payment. Actions wait for your approval by default.

A bounded first request: My open source project [repository URL] helps [user] [get result]. Find 20 recent conversations where [user] describes [problem], with sources, suggest one channel for a launch and draft a Show HN title and a first comment within HN's rules. Do not post or send anything.

Sources, limitations and refresh

The fake-star study, GitHub's Acceptable Use Policies, traffic, topic and social preview documentation, the traffic REST API, the Show HN guidelines, npm and PyPI download documentation, GitHub's maintainer marketing panel and Runa Capital's Q3 2026 ROSS Index were read on November 2, 2026. GitHub does not publish how Trending or search ranks repositories, and download counts include automated installs. The audience and channel tables, checklists, plan and ledger are editorial guidance; the funnel example is illustrative. Nothing here guarantees stars, users or customers.

This guide will be reviewed by February 2, 2027, or sooner if GitHub changes its star, traffic or Trending rules.

Product capability

The Soloop manual describes research with sources, X publishing with OAuth, Reddit preparation with manual publication, cold email through a connected Gmail account and Google Search Ads after review and payment; approval comes first by default. It does not manage GitHub repositories, stars or package releases.

Soloop product manual