Mike Perham writes a background job library for Ruby, gives it away, and sells two closed-source extensions of it. That is the whole business. He has no employees, does all the support himself, and told Rob Walling's podcast in May 2023 that annual revenue was "closer to 10 million than 1 million," from about 2,000 customers. The parts a solo developer can copy are specific: keep the core open so it earns trust, sell the features companies need at scale, price per organization rather than per seat, and never discount. The part you cannot copy is the decade of open source credibility that made the first customers show up on launch day.
The free half and the paid half
Sidekiq launched in early 2012 and took Perham about two months to build. The open source version, still free today, includes scheduled jobs, error handling, and the web dashboard: enough that thousands of Ruby shops run it in production without paying anything.
The paid tiers sell what growing companies need next. Sidekiq Pro costs $99 a month or $995 a year with unlimited usage (checked 3 August 2026) and adds batch processing and stronger reliability guarantees. Sidekiq Enterprise starts at $269 a month per 100 threads and adds rate limiting, cron jobs, unique jobs, and encryption. Both carry a two-week, full money-back guarantee.
Notice what the split does. A free user who scales will eventually lose a job to a Redis restart, want a batch workflow, or need rate limiting. The upgrade is not a sales conversation. It is the same library, from the same author, solving the next problem on the list. In a 2017 interview with Indie Hackers, Perham described the conversion as users who "naturally convert as they find that they need those paid features."
The license is the product
Perham wrote up his method in a 2015 post, How to Charge for your Open Source, and it is blunter than most licensing advice. Publish the commercial code where people can see it if you like, but license it so companies must pay: he suggests AGPL as the default with a purchasable commercial license, since few businesses will accept AGPL terms. "The source may be viewable on GitHub but that doesn't mean anyone can use it for any purpose," he wrote. And the line that should be pinned above every maintainer's desk: "if you don't charge for it, I guarantee you will make no money."
For a small library, his pricing advice is equally concrete: charge roughly your hourly rate per major version. For Sidekiq he chose flat organizational pricing instead, which means a 5-person startup and a 500-person company running one app pay the same. That flatness is what makes a one-person sales process possible: there is nothing to negotiate.
The numbers, dated
Sidekiq Pro went on sale on 1 October 2012, roughly eight months after the open source release. Five to ten customers bought almost immediately, according to the Indie Hackers interview: people who had watched him ship and answer issues for months. He kept his day job about 18 months and quit around 2013, when Sidekiq revenue passed his salary.
By May 2017 he reported about $80,000 a month from roughly 800 customers, growing 50 to 100 percent a year. By May 2023 it was about 2,000 customers and revenue in the high seven figures; a November 2023 interview put it at $7 million. There is no audited figure, but the trajectory across three interviews six years apart is consistent.
What zero employees actually requires
The interesting discipline is everything Perham refuses to do. He gives no discounts, ever, a rule he stated flatly back in 2017. He accepts credit cards only, because purchase orders and invoicing create administrative work that a company of one cannot afford. He does every support ticket himself and treats the tickets as his product roadmap; better error messages and documentation are how he keeps the volume survivable. He hires a designer about once a year for graphics and talks to a lawyer about once a year on licensing. That is the entire org chart.
This is the same trade-off we described in the one-person software company: every process you add either has to be automated or it becomes your job. Perham's version is aggressive because his customers are companies, not consumers. Two thousand businesses paying around $1,000 to $3,000 a year generate far fewer tickets than two hundred thousand consumers paying $10.
Copy this if
The model transfers under three conditions. First, you maintain something businesses already run in production; open core only works when the open part has real adoption, and Perham had spent years on open source and technical blogging before Sidekiq. He calls it ten years to overnight success. Second, the paid features must be scale features (reliability, batching, compliance) rather than core features, or you poison the free tier's trust. Third, you must be able to price per organization at a level that makes 1,000 customers a real income, which usually means pricing higher than feels comfortable.
If those hold, the arithmetic is the appeal. There is no funnel to feed and no team to manage: a boring, stable product, a license, and a price list that has barely changed in a decade. If they do not hold, charging for open source produces what it produces for most maintainers: donations. Perham's point is that the difference is not luck. It is whether you built something companies must pay for, and then asked them to.
Discussion
Sign in with Google or just a name. No email link, no password to remember.