Indie GEO · closed loop

Score → ship → half-life re-score

Profound × Estée Lauder is enterprise GEO: Context Manager as an org brain across a prestige portfolio — brand records, synced context, visibility at scale. Useful, and still open-loop without a graded ship history. Growth Radar is the indie foil: score an opportunity, ship a PR, re-score the metric, then half-life re-check whether the lift still holds — no org-brain Context Manager required. This board is the public proof.

  1. 1 · Score

    Rank the backlog by evidence. Each opportunity carries a metric and a verify condition — not a brand-record vibe.

  2. 2 · Ship

    Open the intervention as a PR on the product repo. The clock starts at merge, not at a Context Manager sync.

  3. 3 · Re-score

    #21 auto-verify grades win / loss / flat. #43 stamps aeo_attribution and soft-fails CRM writeback onto the FeedbackItem.

  4. 4 · Half-life

    #44 re-measures after a second horizon and stamps still_holding + decay — remaining lift ≥ half the verified peak, or decayed.

Profound × Estée Lauder

Enterprise GEO · org-brain Context Manager

Portfolio-scale brand memory and AI visibility tooling. Context Manager remembers who you are so an AI marketer can recommend — monitoring and knowledge, not auto-verify writeback that grades a shipped change, then half-life re-checks whether the lift held.

Growth Radar /wins

Indie closed loop · no Context Manager

Same intervention numbers as the dashboard: baseline → outcome, #43 CRM attribution, and #44 still_holding / decay after a second horizon. Shareable Verified ROI — outcomes that survive half-life, not another enterprise org brain.

Cards below are #21 writebacks with complete ROI; still_holding chips appear after #44 clears a second horizon. Empty means nothing has cleared yet — missing is a value. Features · GEO · MCP

Closed-loop outcomes

33%4/12 measured end to end · 6 inside horizon · 2 unmeasurable
4 wins0 losses0 flat

Outcomes, not monitoring. One job re-scores every play at its own horizon and records a move of 10% or more — in whichever direction it went — against the metric that play was meant to change. Each card names the dates it was measured on. Only plays whose BEFORE number came from an instrument are in this rate: a baseline someone typed is always worse than the measured after-state, so a verdict built on one can never come out anything but a win, and counting those would be scoring our own homework.

Weekly Verified Wins · digest

What closed the loop this week

08 Sept – 15 Sept · 7d

Profound Context Manager is an org brain — brand memory and monitoring-shaped recommendations. This digest is the Forrester AEO answer for the last 7 days: #21 auto-verify grades, #32 FeedbackItem close, #33 before → after share — outcomes, not another monitoring chart.

100%4/4 wins this week
4 wins0 losses0 flat

  • LexiClash

    the mode ItemList advertised /<locale>/brain-training which has never existed (real route is /brain-training-word-games), sameAs listed a CrazyGames page that 404s on 4 surfaces, and imageSizes emitted a w=16 srcset candidate the deployed optimizer rejects with 400

    winbroken links: 4 → 0 · -100%

    Card →

  • Feedback DevTools

    Organization sameAs pointed at a PRIVATE GitHub repo, so it 404d for every anonymous visitor, crawler and LLM on 5 pages plus llms.txt — a sameAs that 404s weakens the entity instead of verifying it

    winbroken links: 5 → 0 · -100%

    Card →

  • IL Apartments

    drizzle bound a JS array as ONE scalar param so ANY($1::uuid[]) always failed — /compare?ids= 500d on every request; and loadCityStats read 0 listings as "no such city" so /city 404d for 4 of 8 top cities linked from every /deals page

    winbroken links: 27 → 0 · -100%

    Card →

  • LexiClash

    removed 122 unconfirmed hreflang annotations across 31 landing pages; 38 of 65 pages claimed alternates that never pointed back

    winCTR: 0.00782669980340517 → 0.014172150908972437 · +81.1%

    Card →

Embed · Verified Wins badge

Drop closed-loop wins on any page

growthradar.app/wins/embed

Same #33 before → after numbers and #36 weekly movers, packaged as an iframe badge. Profound Context Manager is an org brain — brand memory without an embeddable graded outcomes surface. Paste this when a README needs the Forrester AEO answer.

<iframe src="https://growthradar.app/wins/embed" title="Growth Radar Verified Wins" width="360" height="168" loading="lazy" style="border:0;border-radius:2px;overflow:hidden;max-width:100%"></iframe>

Verified Wins · directory

Graded outcomes index

4 cards

Profound Context Manager is an org brain — brand memory and monitoring-shaped recommendations. This directory is the Forrester AEO catalog: every shareable #21/#33 before → after writeback, indexed by project and kind — outcomes you can browse, not another monitoring chart.

100%4/4 instrument-backed wins
4 wins0 losses0 flat
  • internal links3
  • schema1

  • LexiClash

    2 verified · 2W / 0L / 0F

    • removed 122 unconfirmed hreflang annotations across 31 landing pages; 38 of 65 pages claimed alternates that never pointed back

      winschema · CTR: 0.00782669980340517 → 0.014172150908972437 · +81.1%

      Card →

    • the mode ItemList advertised /<locale>/brain-training which has never existed (real route is /brain-training-word-games), sameAs listed a CrazyGames page that 404s on 4 surfaces, and imageSizes emitted a w=16 srcset candidate the deployed optimizer rejects with 400

      wininternal links · broken links: 4 → 0 · -100%

      Card →

  • Feedback DevTools

    1 verified · 1W / 0L / 0F

    • Organization sameAs pointed at a PRIVATE GitHub repo, so it 404d for every anonymous visitor, crawler and LLM on 5 pages plus llms.txt — a sameAs that 404s weakens the entity instead of verifying it

      wininternal links · broken links: 5 → 0 · -100%

      Card →

  • IL Apartments

    1 verified · 1W / 0L / 0F

    • drizzle bound a JS array as ONE scalar param so ANY($1::uuid[]) always failed — /compare?ids= 500d on every request; and loadCityStats read 0 listings as "no such city" so /city 404d for 4 of 8 top cities linked from every /deals page

      wininternal links · broken links: 27 → 0 · -100%

      Card →

Verified ROI · public

Wins you can link

Each card is a #21 auto-verify writeback graded win, loss, or flat — the same Verified ROI #25 shows in-app, with #43 CRM attribution and #44 still_holding / decay after a second horizon. Paste when you need the indie GEO answer: did visibility work move a business metric, and does the lift still hold — without an enterprise Context Manager.

Closed-loop outcomes

33%4/12 measured end to end · 6 inside horizon · 2 unmeasurable
4 wins0 losses0 flat

Outcomes, not monitoring. One job re-scores every play at its own horizon and records a move of 10% or more — in whichever direction it went — against the metric that play was meant to change. Each card names the dates it was measured on. Only plays whose BEFORE number came from an instrument are in this rate: a baseline someone typed is always worse than the measured after-state, so a verdict built on one can never come out anything but a win, and counting those would be scoring our own homework.

FAQ

How is Growth Radar /wins different from Profound × Estée Lauder enterprise GEO?
Profound × Estée Lauder is enterprise GEO with Context Manager as an org brain — brand records and portfolio-scale AI visibility. Growth Radar is the indie closed loop: score an opportunity, ship an intervention PR, re-score after the horizon (#21/#43 CRM attribution), then half-life re-check still_holding / decay (#44) — without an org-brain Context Manager.
What does score → ship → half-life re-score mean?
Score ranks the backlog with a metric and verify condition. Ship opens the fix as a PR; the clock starts at merge. Re-score is #21 auto-verify plus #43 aeo_attribution writeback. Half-life is #44: after a second horizon, re-measure and stamp whether remaining lift is still ≥ half the verified peak.
Are the /wins cards public?
Yes. Each card is a graded auto-verify writeback with metric, baseline, and outcome — the pasteable indie closed-loop foil, not a gated enterprise dashboard. Still_holding / decay chips appear once #44 has run.
What is the weekly Verified Wins digest?
A rolling 7-day rollup of the same public /wins cards — win/loss/flat counts, win rate, and top movers with before → after deltas. Paste /wins/weekly when you need this week's closed-loop answer. Enterprise Context Manager has org-brain memory; it does not publish what graded after ship or survived half-life.
Can I embed a Verified Wins badge on my site?
Yes. /wins/embed is an iframe-ready badge built from the same #33 before → after writebacks and #36 weekly movers — closed-loop win rate plus this week's top mover. Copy the snippet on /wins. Profound's org brain has no embeddable graded-outcomes surface.
What is the Verified Wins directory?
A public index of the same shareable /wins cards — grouped by project, facetable by kind — at /wins/directory. Paste when you need the graded-outcomes catalog. Context Manager holds brand records; it does not publish a browsable directory of what graded after ship and still holds.