PostHog

Openness became an interface, not a constitution

PostHog began in 2020 as a self-hosted, open-source alternative to hosted product-analytics services. By the end of 2025, it presented a much wider platform: analytics, replay, feature delivery and experiments, error tracking, surveys, data movement, and a warehouse. It also published much of the operating manual for the remote company building those products.¹²

Those openings carried real rights. Code outside the ee directory could be inspected, modified, forked, and run under the MIT license; outsiders could propose code and handbook changes; distributed employees could work from a shared written record.³⁴ They did not turn the institution into a member-governed commons. PostHog retained authority over what entered its repositories, which products received investment, which code required a commercial license, who remained employed, and how the corporation raised capital.⁵⁶⁷ The interesting organizational question is therefore not whether PostHog was “open” or “closed,” but which kinds of authority each opening actually carried.

A four-week product arrived with source attached

James Hawkins and Tim Glaser had been testing other startup ideas when they shifted to product analytics during Y Combinator's Winter 2020 batch. Their February 20 launch post offered a self-hosted product intended to keep customer behavioral data from having to flow to another analytics vendor. A repository license fixed three days earlier placed the software under MIT terms.¹

The combination made the product legible and runnable before the company had a long operating record. Developers could inspect implementation choices, deploy the system themselves, report problems in public, and leave with a fork if the maintained product moved elsewhere. The public launch thread also exposed questions about pricing, hosting, data control, and the viability of an open-source business directly beside the founders' account. Visibility did not settle those questions, but it let prospective users contest them before buying.¹⁵

The license boundary changed within the first year. An August 2020 repository commit introduced an ee directory with a separate enterprise license while leaving code outside it under MIT. That division remained visible at the December 31, 2025 repository revision: the root license granted ordinary MIT rights outside ee, while the enterprise license restricted production use to customers with a valid commercial agreement.³ “Open source” therefore described a substantial core and a route to self-hosting, not every feature or every production use.

Independent reporting in June 2021 described PostHog as remote-first and open-core, described its product-analytics, feature-flag, and session-replay offerings, and reported a $15 million Series B led by Y Combinator's Continuity Fund and GV. The report also distinguished the MIT product from differently licensed and hosted plans rather than treating them as one undifferentiated offering.⁸ Capital, community participation, and commercial control were developing together; the available record does not isolate which one caused adoption or growth.

The point tool widened into a product and data platform

PostHog's own retrospective says the product already combined analytics, feature flags, and session replay by November 2020. In 2023 the company began describing it as a “Product OS,” ended support for a paid self-hosted deployment, emphasized its cloud and open-source projects, and chose to build a warehouse after, it says, a large customer made another warehouse its source of truth. By 2024 it reported organizing several products as distinct businesses joined by common data and infrastructure.⁹

The repository at the end of 2025 makes that widening inspectable without requiring the company's adoption or revenue claims. Its README lists product and web analytics, replay, flags, experiments, error tracking, surveys, a warehouse, and data pipelines, among other products. The same page recommends PostHog Cloud as the fastest route and describes open-source self-hosting as a hobby deployment without company support or guarantees.²

Consolidation can remove handoffs among separate tools and give one product team more shared context. It can also move more telemetry, release controls, customer records, and operating dependency behind one vendor's interfaces. PostHog's strategy explains why its teams wanted that concentration. Comparative evidence on customer privacy, portability, reliability, switching cost, and product outcomes is not available here.

The handbook turned remote coordination into editable text

PostHog's retrospective gives its handbook an unexpectedly tactical origin: the founders wanted a young company to look mature at launch, so they wrote down how they intended to work. The company says the exercise then changed how it operated because employees and outsiders could propose edits and public writing forced decisions into clearer form. It names GitLab as an inspiration for transparency while emphasizing that PostHog adopted a different business and operating model.¹⁰

By the end of 2025, the public handbook described asynchronous communication as the starting point for colleagues across countries. Work was to begin in issues and pull requests when possible, with no expectation of replies outside planned hours. The same policy drew a visible boundary around transparency: product, roadmap, handbook, and strategy were generally public; financial performance, security, fundraising, recruiting, customer data, compensation, and disciplinary matters moved into internal or private systems.⁴

That boundary matters. A public manual makes formal expectations inspectable and gives a remote worker or applicant something concrete to question. It does not show whether people experience the rules as promised, whose edits are accepted, how informal influence travels in private channels, or whether workers can challenge consequential decisions without cost. Independent worker evidence on those questions remains absent.

Small teams owned products inside a centrally chosen portfolio

PostHog's handbook divided much work among teams of two to six people. It assigned each team a customer and an area, made the team responsible for usage, quality, and revenue, and gave it the final call on features within its product and the existing release system. Teams were also expected to speak with users, run weekly retrospectives and sprints, maintain a roadmap, and monitor product metrics.¹¹

The delegation stopped at a stated frontier. The executive team set company direction, selected products, made key hiring, pay, and disciplinary decisions, and coordinated cross-team matters such as pricing. The handbook said Hawkins and Glaser focused on product sequencing. Another page invited anyone to propose or prototype a new product, but required the central Blitzscale team to approve the sustained investment needed to build it.⁶

This was not simply decentralization. It was nested authority:

• an engineer or outside contributor could propose a change;
• an owning team could decide features within an existing product;
• executives chose the portfolio and controlled employment and cross-team commitments;
• corporate capital and control sat outside those teams in arrangements not disclosed in the handbook.

The structure may let small groups move with fewer approvals while keeping a coherent platform. Public rules alone cannot establish speed, product quality, worker autonomy, or the absence of shadow hierarchy.

The open door still had owners

The end-2025 contribution guide welcomed external pull requests but made the boundary unusually plain. Owning teams triaged submissions; support work could delay review; maintainers could close work that was out of scope, stale, or costly to maintain; paid features were generally not open to external contribution; and roadmap work was often already assigned internally.⁵ Fork rights gave outsiders an alternative to acceptance, but a fork is not the same as a vote over the maintained product, cloud service, trademark, customer relationships, or company budget.

PostHog's own account of corporate control also became more explicit in 2025. In its Series D announcement, the company called itself founder-controlled and said an earlier financing had increased founder control while providing some employee liquidity. It reported $70 million of new primary capital led by Stripe at a $920 million valuation. A later Economic Times report, citing Hawkins's LinkedIn post, described a $75 million funding round led by Peak XV at a $1.4 billion valuation. A Florida annual report filed in April 2025 identifies PostHog Inc. as a foreign profit corporation and lists Hawkins as chief executive and Charles Cook as chief operating officer.⁷ These records establish corporate form, offices, financing events, and PostHog's own control claim—not the undisclosed cap table, board composition, investor protections, voting thresholds, or the day-to-day influence of any investor.

PostHog makes a useful companion to the Linux ecosystem: both expose code and invite participation, but PostHog joins those practices to a single venture-backed product company and hosted platform. Its lasting question is precise. Publication can make authority easier to inspect, and licenses can make some capabilities portable, without making authority equal. The next research needs to follow the people who use those openings: external contributors whose work was accepted or refused, customers who migrated in or out, and workers across countries who encountered the public operating system from inside it.