Cohortica

A free, open-source interview-prep site for marketing roles, and a content-architecture problem disguised as a web app. Cohortica organizes 59 practitioner playbooks across 12 disciplines (email marketing, paid search, SEO, paid and organic social, paid media, web and customer analytics, attribution and experiments, competitive research, GEO and LLM traffic, and product management), each one condensing a topic into the question it answers, the framework a practitioner actually uses, the metrics that matter and their traps, and three to five realistic interview questions with the reasoning a strong candidate would use. Two deterministic diagnostics, a funnel engine and a brand strategy engine, sit on top and train the same diagnostic thinking a case interview tests for. No AI, no backend, no accounts, no tracking; the whole thing is static and runs client-side.

Why I built it

I chose to study marketing at Syracuse University’s Whitman School of Management because of its emphasis on product management and marketing analytics, the two disciplines that frame Cohortica’s playbooks. That training changed how I saw my own work. Looking at documentation as a product manager would, I realized how far technical communication still had to go in the companies I had worked for: it was treated as an afterthought to shipping, not as part of what was shipped.

That realization is why I am joining Canonical: to learn from the teams that set the standards in this field, and to make the case that documentation is a product feature. Open source makes software available; documentation is what makes that availability usable. For an open-source data solution, the readers are different teams: data engineers, data-platform operators, data analysts and scientists, and compliance teams. Each one has to adopt the software, use it in the way that gets the most value from it, troubleshoot it, and extend or adjust it to fit their own stack. Serving all of them does not mean writing separate descriptions of the same product. It means giving each team its own path through one governed body of knowledge. Cohortica is a small, self-contained exercise in exactly that: one fixed structure, written once, that readers from twelve disciplines can each find their way through.

Why this is a documentation project, not just a product

The interesting design decision was never the diagnostics’ logic; it was deciding that a playbook is a fixed shape, not free-form writing. Every one of the 59 follows the same four movements (an opening that earns the reader’s next minute, the framework, the metrics with their traps, and the interview questions with the reasoning behind them), and that shape is enforced editorially, not by code. The result reads like one voice wrote all 59, even though the goal was to make each one fast to extend without a style drift creeping in over time.

The two diagnostics apply the same discipline to decision logic instead of prose: each is one data array of states (declining, flat, inefficient for the funnel; absent, weak, repositioning, and similar triads for each brand phase), and every state carries the same fields (signals, actions, what to verify, and a linked micro-playbook). Adding a new state or a new discipline means filling in that shape once; the app reads the data generically and needs no code changes to grow.

Open source, with real contribution guidelines

Cohortica is MIT-licensed and open to contributions, and the highest-value one is a playbook: drop a markdown file with three frontmatter keys into the right discipline folder, and the app picks it up at build time. The repository’s README doubles as the full contributor guide (there is no separate documentation site for a project this size), covering how to add a playbook, add a new discipline, and extend either diagnostic; its CONTRIBUTING guide carries the editorial bar content contributions are held to: the four-movement format, the tone rules (an operator’s voice, benchmarks over adjectives, no product pitches, no named clients), and the review checklist every content pull request runs through.