Diagnose your growth engine in 3 minutes. Two free audits on one shared engine — the AARRR Growth Health Score and a PLG Readiness Score — each giving you a 0–100 score per stage, your top bottlenecks, an ICE-prioritized experiment roadmap, and an exportable action playbook.
Built by GrowthHackers — free and open-source forever. Zero backend, privacy-first: no database, no PII ever leaves the browser.
Live demo: — deploy your own in one click ↓
Screenshots are real and reproducible:
npm run build && npm run screenshotregeneratesdocs/screenshot.png,docs/results.png, anddocs/embed.png.
- Two diagnostics, one engine — switch between the AARRR growth audit and a PLG Readiness audit (Time-to-Value, Self-Serve, Product-Qualified Leads, In-Product Virality, Expansion-Led Revenue). Both fully selectable.
- 0–100 score per stage — weighted by growth-model importance, with a clear overall score and label.
- Bottleneck detection + ICE experiments — your 3 weakest stages, plus the single highest-ICE experiment from each.
- Actionable playbook — per-weak-stage tactics with concrete steps, the metric to watch, effort, and horizon. Exportable as Markdown (and in the PDF).
- "You vs. median" benchmarks — per-stage, switchable across 5 industries (All · B2B SaaS · PLG SaaS · Marketplace · Fintech).
- Built-in growth loop — shareable result link, auto-generated social preview, PNG and PDF report export.
- Embeddable — drop the audit on any site with one
<script>tag. - 9 languages via
?lang=(EN, AR-RTL, IT, NL, ZH, ES, FR, DE, PT-BR).
Every result is shareable, and shared links preview the score on social — pulling new people in to take their own audit.
- Shareable link, no PII. Your result is encoded entirely in the URL. AARRR
uses
?r=1.<15 digits>; any diagnostic uses the self-describing?r=2.<id>.<digits>(e.g.?r=2.plg.342013402230114). That's the entire payload — one answer value (0–4) per question, no email, name, or timestamp. Opening the link recomputes the exact score and breakdown. - Dynamic share image.
app/api/ogrenders a branded card from the token vianext/og, so a pasted link previews the score. Stateless — nothing stored. - Ungated for recipients. A shared link shows the full breakdown immediately, with a CTA to take your own audit.
- PNG + PDF report. One click saves a branded PNG, or a full multi-section PDF generated entirely client-side (score, breakdown vs. median, bottlenecks + recommendations, the playbook, top experiments).
One script tag, self-contained, no backend (full docs →):
<script
src="https://growth-score.growthackers.io/embed.js"
data-diagnostic="aarrr"
async
></script>It injects a responsive iframe and auto-resizes it. Options: data-diagnostic
(aarrr / plg), data-lang, data-token (embed a specific result),
data-height, data-base.
git clone https://github.com/aymandakir-gh/gh-growth-score.git
cd gh-growth-score && npm install && npm run devOpen http://localhost:3000. No env vars required — the
app runs fully out of the box. Try ?d=plg, ?lang=ar, or ?industry=fintech.
Everything is a pure function on the shared engine in
lib/engine.ts. A diagnostic is just data — a Diagnostic
descriptor (dimensions + questions + experiments). AARRR lives in
lib/scoring.ts and PLG in
lib/diagnostics/plg.ts; both run on the same engine,
share codec, OG/report, PDF, and playbook pipeline.
1. Answer the questions. Each question has a 5-point scale, scored 0
(critical gap) to 4 (best-in-class).
2. Score each stage (0–100): round( sum(answers) / (questions × 4) × 100 ).
3. Weight into an overall score. AARRR weights (sum to 1.0):
| Stage | Weight | PLG dimension | Weight | |
|---|---|---|---|---|
| 🎯 Acquisition | 20% | ⏱️ Time-to-Value | 25% | |
| ⚡ Activation | 20% | 🛒 Self-Serve Funnel | 25% | |
| 🔄 Retention | 25% | 🎯 Product-Qualified Leads | 15% | |
| 💰 Revenue | 25% | 🔗 In-Product Virality | 15% | |
| 📣 Referral | 10% | 📈 Expansion-Led Revenue | 20% |
4. Find the bottlenecks (3 lowest stages) and prescribe experiments — the
highest-ICE experiment from each, where ICE = round(Impact × Confidence ÷ Effort).
5. Compare + recommend + plan. Each stage is compared to an industry median
(lib/benchmarks.ts), each bottleneck gets a tailored
recommendation, and the playbook (lib/playbook.ts)
gives concrete tactics.
Benchmark medians (
datasets/benchmarks.md) and the playbook (datasets/playbooks.md) are honestly-labelled synthesized/editorial content, not licensed data. Swap in your own and the labelling notes how.
The OSS build keeps the email gate but makes it fully optional, so the tool works with zero backend:
- Submitting
POSTs toapp/api/lead/route.ts, a server-side proxy that keepsLEADS_API_URLprivate. If set, the lead is forwarded; if unset (default), the route logs the score only (no PII) and returnsokso results still unlock. - Rate-limited to 10 req/min/IP. Shared links and the embed are ungated.
Turnkey — vercel.json is { "framework": "nextjs" }, no required secrets. From
a clone, one command ships it:
npx vercel --prodOptionally add LEADS_API_URL, NEXT_PUBLIC_SITE_URL, NEXT_PUBLIC_SENTRY_DSN,
NEXT_PUBLIC_POSTHOG_KEY in Project Settings → Environment Variables. All
optional — the app degrades gracefully when absent.
npm test # 223 unit tests (Vitest)
npm run e2e # 33 Playwright e2e (needs a build) — 256 tests total
npm run lint
npm run build
npm run screenshotCI (.github/workflows/ci.yml) runs lint → test →
build → e2e on every push/PR to main.
- Unit: the shared engine + AARRR parity, both diagnostics, the codec
(v1/v2/legacy), benchmarks + per-industry depth, recommendations, playbook,
PDF, share-model,
/api/lead, rate limiter, i18n +resolveLocale, embed helpers,LanguageSelector. - E2E (Playwright + axe): the share loop (AARRR + PLG), OG/PNG/PDF export,
the playbook export, the embeddable widget (real
embed.jsiframe + auto-resize), i18n (?lang=+ Arabic RTL), the industry switch, and axe accessibility scans (WCAG 2.0/2.1 A & AA) on landing, both shared results, the embed, and the RTL page — all clean. - Lighthouse (desktop, committed under
docs/lighthouse/): Performance 100 · Accessibility 100 · Best Practices 96 · SEO 100.
- Next.js 14 (App Router) · TypeScript · Tailwind CSS
- jsPDF client-side PDF ·
next/ogdynamic share images (Edge) - Vitest unit · Playwright e2e + axe a11y + screenshots
- Sentry + PostHog observability (optional, graceful degrade)
- No external UI libraries — pure Tailwind components
app/
page.tsx # Server component — diagnostic-aware generateMetadata
embed/page.tsx # Embeddable widget surface (noindex)
api/lead/route.ts # Lead capture proxy (validation + rate limit + fallback)
api/og/route.tsx # Dynamic share card (next/og, Edge) — diagnostic + industry aware
lib/
engine.ts # Generic pure scoring core — every diagnostic runs on this
scoring.ts # AARRR diagnostic (descriptor + back-compat API) → delegates to engine
diagnostics/ # plg.ts (PLG Readiness) + index.ts (registry + share codec)
benchmarks.ts # Per-diagnostic × per-industry medians + comparison
recommendations.ts # Per-dimension tailored advice by band
playbook.ts # Band-aware tactics + Markdown export
pdf-report.ts # Client-side PDF (jsPDF), pure → node-testable
share-model.ts # Pure view-model for OG/PDF (diagnostic + industry aware)
embed.ts # Pure embed-loader helpers (mirrored by public/embed.js)
i18n.ts / i18n-context.tsx # 9 locales, ?lang=-driven (never localStorage)
components/ # HomeClient, GrowthQuiz, ResultsDashboard, PlaybookSection, ShareCard, …
public/embed.js # Self-contained embeddable-widget loader
datasets/ # benchmarks.md + playbooks.md (documented, honestly-labelled)
e2e/ # Playwright specs (loop, diagnostics, playbook, pdf, embed, i18n, benchmarks, a11y)
A diagnostic is data. Add a Diagnostic descriptor (dimensions + questions +
experiments) and register it in lib/diagnostics/index.ts. The engine, codec,
OG/report, PDF, playbook, benchmarks, and i18n all pick it up. A descriptor
integrity test (lib/engine.test.ts) enforces weights sum to 1, 5 options per
question, unique ids, and a non-empty experiment bank per dimension.
MIT — use it, fork it, white-label it. Attribution appreciated, not required.
Built with care by GrowthHackers.