Open-source the tool, then sell the hosted version

Open-sourcing the tool and then selling the hosted version means you give the code away and charge the teams who would rather not run it themselves. Sentry started as an open-source project and later launched a paid cloud version that had a paying customer on day one. SimpleLogin opened all of its code from the start and sells on trust, but publishes no revenue figures.

How it works

1. Release code that solves your own problem. Sentry's founders built it to fix their own mistakes, as the Sentry about page explains. It is now an error and performance monitoring platform for developers.
2. Let engineers carry it between jobs. Cramer's bet was that engineers who had used Sentry at Uber or Airbnb would bring it to their next company, and that those younger startups would pay for the convenience of hosted.
3. Ship a thin hosted version with billing. In an interview on Sentry's path to product-market fit, Cramer describes building a cloud service, adding Stripe billing and launching it over Christmas break. He ran it as a side experiment while still working at Disqus.
4. Price low to test demand. Sentry had one paying customer on day one and ten within a few days. Every one of them arrived organically, at seven dollars. See the Sentry case for the full timeline.
5. Use open code as the trust argument. The SimpleLogin homepage says every component is open source, including the server, browser extension, JS library and mobile apps. It sets this against handing your email to a black box. The same code lets anyone run SimpleLogin on their own server. See the SimpleLogin case.

The open project is the top of a growth loop: usage spreads, some users change jobs, and some of those teams pay for hosting. Cramer says Sentry aimed at a long-tail funnel that would play out over 10 years, and that this worked very well for them. Sentry later reported thousands of paying customers, $600K in annual revenue and profitability.

What it takes

Our estimate: about 80 hours of work, and about 30 days before you can read a signal. Most of the hours go into the hosted service itself: multi-tenant setup, signup, billing, basic monitoring and a pricing page. The open-source project should already exist, because building an audience for the code takes far longer than this. A month gives organic traffic from the repo and docs time to reach the paid page and produce a handful of decisions.

What to measure

Compare paid hosted signups with new self-host installs over the same weeks. A healthy reading is hosted growing alongside self-host. Self-host growing while hosted stays flat suggests the convenience gap is too small.

Track where paying customers come from. Sentry's early buyers were all organic. If your first payers only come from ads, the open project is not doing the work.

Measure conversion among teams that ran the open code first. This is the core bet in Sentry's plan. Our free-to-paid conversion benchmarks give context, though open-source funnels differ from freemium.

Read hosted retention by monthly cohort. Later cohorts that churn faster than early ones often mean early adopters were fans, not typical buyers. Finally, check revenue per customer against hosting cost, because a low launch price can hide negative margins.

When it backfires

• Self-hosting is easy enough. SimpleLogin markets that you can run it on your own server. That is good for trust, but if your tool is just as easy to run yourself, few will pay for hosting.
• Your users are not short on ops time. Sentry's plan depended on younger startups choosing convenience. Buyers with spare infrastructure staff may self-host forever.
• You expect fast payback. Cramer describes a funnel built over a decade. If you need revenue this quarter, this tactic is slow. Check your payback period assumptions.
• You read reach as revenue. SimpleLogin says it has helped over 50,000 people since its 2020 launch. That figure does not say how many pay.
• The launch price sticks. Seven dollars validated demand for Sentry. Our view is that you should plan a price review before the low price becomes the expectation.

Run it as an experiment

We recommend this hypothesis: "Users of our open-source project will pay for a hosted version, and at least some will arrive without paid acquisition." A clean split is hard when you have one hosted offer, so we suggest a before-and-after comparison with a fixed 30-day window, plus a split test on the pricing page if traffic allows. Size it with the A/B test sample size calculator.

Log the signup source, whether the account ran the open code before, the plan chosen, the first payment date, hosting cost per account, and self-host install counts for the same days. Our recommended decision rule: keep going if paying customers arrive organically and hosted signups grow with installs. Rework the convenience gap or price if installs grow and paid stays flat after the window. Follow the steps in how to run a growth experiment and track follow-ups in the growth experiment backlog template.

How strong the evidence is

The grade is case with numbers: at least one company documented the tactic and published results as numbers. Sentry carries the grade, through Cramer describing the launch and its results on the record. The evidence has no baseline and no conversion rate from open-source users to payers. The $600K and profitability figures come later than the launch and are not split by hosted versus other revenue. Sentry's current reach of millions of developers and 200K organizations across 146 countries is not attributed to this move. SimpleLogin is a mixed case. The SimpleLogin about page confirms the code was open from the 2019 start and drew community contributions. Its user figure is self-reported and comes with no revenue data.