# PDR 004: Build a product-led growth learning loop

- HTML version: https://robbiepalmer.me/projects/personal-knowledge-graph/pdrs/004-product-led-growth-learning-loop
- Project: Personal Knowledge Graph
- Status: Accepted
- Opened: 2026-02-14
- Decided: 2026-02-14

## Context

Publishing a page or feature proves delivery. It does not show whether people
discover it, understand it, return to it, or use it to reach related work. Page
counts alone also miss the paths that matter, such as moving from an idea to a
project, following a decision to its evidence, or trying an incubated product.

The Personal Knowledge Graph is both a public body of work and an incubator.
Usage can reveal where onboarding fails, which connected paths help readers,
and which projects attract enough repeated interest to justify more investment.

## Decision

Use product-led growth loops to improve discovery, activation, engagement, and
return use across the public graph and its incubated products.

Start each measurement with a product question. Record the smallest useful
event, review the result, change the content or product, and measure the next
version. Use experiments when competing presentations have a clear outcome to
compare. Treat observed behaviour as evidence for a decision, never as an
automatic priority score.

Use the same loop to find productisation candidates. Repeated use, deeper
journeys, return visits, and explicit follow-up interest can justify a separate
product decision, customer research, or a financing case. Traffic by itself
does not establish demand or a business model.

Keep collection proportionate. Do not collect names, email addresses,
advertising profiles, or unrestricted session data merely to enrich a
dashboard. A more invasive collection method requires a separate product
decision.

## Consequences

Content and feature work can respond to observed use instead of stopping at
publication. Promising projects gain evidence for further research,
productisation, and possible financing.

The loop can reward easy-to-measure activity over long-term value. Client-side
events are incomplete, and a small audience makes many comparisons noisy.
Product judgment and direct conversations remain necessary.

## Evaluation

Each retained event must answer a named question and lead to a review or
decision. Revisit the policy if events accumulate without changing the product,
if experiments chase shallow engagement, or if privacy costs outweigh what the
team learns.

## Migration note

This record extracts the product-led learning loop and privacy policy from ADR
040\. It preserves the Accepted status and 2026-02-14 date. ADR 040 retains the
technical selection and integration of PostHog.

## Related records

- Project: [Personal Knowledge Graph](https://robbiepalmer.me/projects/personal-knowledge-graph.md)
- Implemented by ADR: [ADR 040: PostHog Analytics Platform](https://robbiepalmer.me/projects/personal-knowledge-graph/adrs/040-posthog-analytics.md)

---

Markdown index of this site: https://robbiepalmer.me/llms.txt