Growth

Product Feedback to Measure Loop

Turn feedback + analytics into one measured product change per cycle; keep bugs separate.

Use when

Customer feedback and product telemetry are piling up, but shipping is driven by loudest request instead of a measured build-learn cycle.

Advanceddifficulty
Growthcategory
Greg Isenberg + Ellie interview: Making $$$ with Loop Engineering — ultimate product feedback loop (feedback + analytics + logs → prioritize → prototype → measure); split from bug/uptime loopssource

Cadence

Weekly product tick; bug/uptime work stays in a separate reliability loop

Verification

Top pain themes are ranked with evidence, one feature or UX change is prototyped or shipped behind a flag, a pre-chosen KPI (activation, retention, NPS, revenue proxy) is measured after the window, and bug/uptime work was not mixed into this cycle.

Structured loop spec

FieldValue
NameProduct Feedback to Measure Loop
CategoryGrowth
TriggerWeekly product tick; bug/uptime work stays in a separate reliability loop
ObjectiveTurn feedback + analytics into one measured product change per cycle; keep bugs separate.
Allowed inputsRelevant files, source notes, logs, tests, screenshots, metrics, or task state for this loop
Allowed actionsIngest feedback sources (support, sales notes, reviews) plus product analytics and error trends. Tag items as bug vs feature vs noise.; Route pure reliability items out to a bug/uptime loop (error rate, MTTR, Sentry) — do not let them steal this cycle's feature budget unless they block the KPI.; Cluster remaining pains, rank by frequency × severity × strategic fit, and pick one change with an explicit success KPI and stop condition.; Prototype or ship the smallest reversible version (feature flag preferred). One change per round.; After the measurement window, compare the KPI to baseline, log keep/iterate/revert, and update the product experiment log.; Require human approval for user-visible launches, pricing, data-model changes, or anything that emails customers. Stop when the chosen KPI moves, two cycles show no learning, or risk requires a human gate.
VerificationTop pain themes are ranked with evidence, one feature or UX change is prototyped or shipped behind a flag, a pre-chosen KPI (activation, retention, NPS, revenue proxy) is measured after the window, and bug/uptime work was not mixed into this cycle.
Stop conditionStop when the verifier passes, the budget is exhausted, no progress is made, a blocker appears, or approval is required.
BudgetSet a time, turn, token, retry, file, or dollar cap before running the loop.
Approval boundaryHuman approval required before publishing, sending, deleting, spending, changing accounts, touching production, or making reputational/legal/financial commitments.
Safe outputDraft, report, checklist, table, or approval-gated recommendation
Works withClaude, ChatGPT, Gemini, any tool-using AI assistant

Steps

  1. Ingest feedback sources (support, sales notes, reviews) plus product analytics and error trends. Tag items as bug vs feature vs noise.
  2. Route pure reliability items out to a bug/uptime loop (error rate, MTTR, Sentry) — do not let them steal this cycle's feature budget unless they block the KPI.
  3. Cluster remaining pains, rank by frequency × severity × strategic fit, and pick one change with an explicit success KPI and stop condition.
  4. Prototype or ship the smallest reversible version (feature flag preferred). One change per round.
  5. After the measurement window, compare the KPI to baseline, log keep/iterate/revert, and update the product experiment log.
  6. Require human approval for user-visible launches, pricing, data-model changes, or anything that emails customers. Stop when the chosen KPI moves, two cycles show no learning, or risk requires a human gate.

Prompt

Run the Product Feedback to Measure Loop. Pull recent customer feedback, product analytics, and error signals. Separate bugs/uptime into a reliability track — only escalate a bug into this loop if it blocks the product KPI. Cluster feature pains, pick one highest-value change, define the success metric and measurement window, and ship the smallest reversible version (flag if possible). After the window, report keep/iterate/revert with numbers, not vibes. Do not email users, change pricing, or merge risky schema changes without approval. Stop on KPI movement, two no-learning cycles, budget, or red approval boundaries.

Run in Claude Code

Paste this into Claude Code (or any tool-using agent) to run the loop bounded: one change per round, the same verification every round, durable state files, and explicit stop conditions.

Run the "Product Feedback to Measure Loop" loop from AI Loop Library (https://ailooplibrary.com/loops/product-feedback-to-measure-loop/) as a bounded loop.
Goal: Turn feedback + analytics into one measured product change per cycle; keep bugs separate.
Rules: one change per round; run the same verification every round (Top pain themes are ranked with evidence, one feature or UX change is prototyped or shipped behind a flag, a pre-chosen KPI (activation, retention, NPS, revenue proxy) is measured after the window, and bug/uptime work was not mixed into this cycle.); append each round to docs/loops/product-feedback-to-measure-loop/progress.md and update docs/loops/product-feedback-to-measure-loop/state.json; stop on verifier pass, 8 rounds, 3 consecutive failed verifications, no progress, a blocker, or anything needing human approval (money, production, outbound, deletion). Finish with a proof report: rounds used, changes made, verification output, remaining risk, and the next human decision.

Get the MCP server + agent pack

Tags

productfeedbackanalyticsPostHogSentryprioritizationapproval gated

Related

Back to all examples