01Research
* Product Evidence Library
Credible product research, translated into clear decisions and copy-paste prompts for the app you're building.
01Research
02Decision
03Implement
Our experience
The problem
“Best practices” spread because they sound right — not because anyone traced the evidence. That turns important product decisions into expensive guesses.
The usual move
The Science-Based Software move
How Science-Based Software works
We do the research work once, then package the result so you can use it while you're actually building.
Start with primary sources and credible research — not recycled “best practices” lists.
Papers · Audits · Telemetry
Turn the finding into a decision: what matters, for whom, and when it applies.
Plain words · Boundaries · Grades
Use a ready-to-paste prompt to apply the principle to your own app or workflow.
Prompt · Paste · Ship
Product evidence
Real screens from real products, audited against the real research.
Evidence library
Each topic pairs the peer-reviewed research with our own audit of what top apps actually ship — then ends with a copy-paste prompt that applies it to your app.
The evidence standard
A recommendation should tell you how much confidence it deserves — not merely how confidently it was written.
Direct, credible evidence supports the claim.
Promising evidence, but context changes the result.
The evidence does not support the common advice.
Science-Based Software playbooks
Build-order checklists with copy-paste AI prompts. Each decision comes graded, cited, and ready to ship — four playbooks, twenty steps, one purchase.
Featured playbook
Every decision cites its evidence
Paste one prompt into Claude Code, Codex, or Cursor. Your agent runs 38 evidence-graded checks on your actual codebase and writes a report with the research behind every fix.
Free · annualSeven questions 2026 actually answered — every number verified or cut, and the six famous "facts" we refused to print named on their own page.
Built by a product manager who ships