How we decide things

Research

Most "data-driven" companies cherry-pick a single example to justify what they already wanted to do. We try not to.

The methodology

For every meaningful decision (anything affecting business model, tech stack, or core feature):

• Minimum 3 working examples, companies/products successfully using this approach, with metrics where possible
• Minimum 3 failing examples, companies/products that tried this and failed, with reasons documented
• Pros/cons matrix, at least 5 pros, 5 cons, weighted by importance to our context
• Counter-evidence search, actively look for sources that contradict the desired conclusion
• DMAI (Define-Measure-Analyze-Improve), what metric tells us if it's working, what threshold triggers a pivot

Anti-patterns we avoid

• "Company X does this and they're successful" → not enough. Find 2 more.
• "This feels right" → flag as [FEELING-DRIVEN] or research it
• Cherry-picking, actively seek disconfirming evidence
• "First strong example", wait for the third example before concluding

Worked example

When deciding "should we be free for consumers and paid for B2B?":

1. Found 3+ working examples (Navan, Replit, Slack)
2. Found 3+ failing examples (Equals, xqa.io, Joovlin)
3. Found disconfirming evidence (Andrew Chen's work on incentivized users)
4. Built pros/cons matrix across 4 options
5. Defined DMAI: free→paid conversion ≥5% by 90 days, support burden <10 tickets per 1000 free users/month, etc.
6. Picked Hybrid (Option B), not Pure Free (Option A), even though Option A felt better, the data said it was riskier at indie scale

What we publish

• The decisions log excerpt at /decisions
• Build progress at /building
• Real spend + traction stats at /stats
• Affiliate disclosures at /affiliates
• Public version history at /changelog

Source code is private during early build. The strategic docs (decisions, research, validation evidence, business model analysis) are summarized on this site rather than exposed verbatim, since they contain partner pricing data and product roadmap details that aren't appropriate for full public release yet.