Canonical source: docs/claude/growth-signals-playbook.md. This page is generated by docs/scripts/sync-handbook.mjs. Edit the source file in the repo; changes appear here on the next build.
Growth Signals Playbook — Signal Architecture + Intent Signals
Distilled 2026-07-12 from two practitioner writeups George supplied (0→7-figures-in-6-months "Signal Architecture"; top-1%-Smartlead "17 intent signals"), as part of the agentic-fleet scoping alongside the Viktor/Zeely/Polsia research.
DUAL-TRACK BY DESIGN (George, 2026-07-12): track (A) is not secondary. The SAME engine serves (A) Ready Practice's own scaling — acquiring clinic clients (B2B), run FIRST as customer-zero — and (B) the clinic-facing Growth Suite (clinics acquiring patients), which productizes the proven version. Performance-based mechanics likewise cut both ways: RP charges clinics per-outcome (booking-funnel attribution edge) AND RP scales itself on performance discipline (PCPL-measured outbound, CAC-payback-gated spend, 30%-recurring affiliate comp) — with the operational learning coming from our own client data first.
1. Signal Architecture — the 5 components (build in this order)
The core claim, which matches our experience: timing + relevance beat personalization. A signal = a public, dated event indicating purchase intent. The architecture is what turns scattered signals into pipeline; signals without it are dashboards nobody acts on.
| # | Component | What it means for Ready Practice |
|---|---|---|
| 1 | Single source of truth — every signal lands in one system with standard metadata (source, timestamp, account, category) | We already have the pattern: the Atlas Intake Router is exactly this for eng signals. Growth needs its sibling: one Firestore collection (growth_signals) every watcher writes into. NOT another dashboard. |
| 2 | Normalized intent scoring — every action gets a point value (blog +2, pricing +8, demo-form +15, usage-spike +20) rolled up per account | Score at the account (clinic) level on the growth CRM lead docs. Start with guessed weights; component 5 fixes them. |
| 3 | Enrichment — merge behavior with firmographics (size, stage, stack) | Clay-equivalent: our lead auto-research (Viktor idea #1) — website/providers/specialties/size/EMR stack → the 1-page brief. Same job. |
| 4 | If/then plays — the system ACTS: Slack alert on threshold, outbound sequence, nurture, retargeting | The part everyone skips. Maps to our heartbeat/worker pattern. Concrete first play (from the article): pre-call enrichment — booked demo → enrich → prep doc in Slack before the call. |
| 5 | Self-correcting loop — closed-won data re-weights the scoring quarterly | Identical to our continual-learning pattern (confirm-card signals, build-agent ledger). Requires #178 (identity stitch / attribution) to know which signals were present on closed deals. |
Key dependency: component 5 (and honest scoring at all) needs #178 attribution foundation. Build the SSoT + scoring first, wire the loop when #178 lands.
2. The 17 intent signals (B2B — Ready Practice selling to clinics)
Metric that matters: PCPL (prospects contacted per lead). Cold list ≈ 500+; signal-based drops it 10x.
Market discovery (find new accounts)
- Running lots of ads (Meta Ad Library is free/public) → they have acquisition budget + care about lead flow → warm target for "convert that spend into booked clients"
- Revealing hires — front-desk/ops/marketing roles at clinics = growth mode; "EMR experience required" = stack in play
- Engagement on the right LinkedIn posts (competitor lead-magnets, clinic-growth influencers)
- Conference exhibitors/sponsors (longevity/med-spa/wellness events) — paid to be in-market
- Competitor/integration mentions on their site (Healthie, Practice Better, Cerbo, Profi)
- Tech stack (running a competitor or complementary tool)
- New decision maker (new practice manager/owner ≤60d — inherited vendors up for review)
- Website traffic surge or crash (momentum or pain — both are openers)
- Niche directories/associations (registries of longevity clinics, med-spa associations) — tight segment, high relevance
- Own-website visitors deanonymized (RB2B/Vector-class pixel on readypractice site) — pricing-page visitors are the hottest list we can get
Account monitoring / ABM (watch a named list)
- Target account posts a revealing job
- Target account changes tech stack
- Target account starts an EMR/system migration
- Target account pursues a compliance cert (HIPAA/SOC2 language in postings)
- Named person complains about a competitor publicly (social listening)
- Stated initiatives in filings (enterprise only — mostly N/A for clinics; applies to franchise/roll-up groups)
- RFP/RFI posted (gov/public wellness contracts; franchise groups)
The two motions underneath all 17
- Database filters (Prospeo-class: executive events, competitor mentions, traffic trends)
- Scrape-and-enrich we build ourselves (Apify/Firecrawl-class + our own agents: job boards, ad libraries, exhibitor pages, directories). Our market-intel engine already does this motion for competitors — the growth version points the same machinery at PROSPECTS.
Freshness is the whole game — a stale signal isn't a signal. Watchers must run on schedules (heartbeat), not one-off scrapes.
3. What this changes for us (decided 2026-07-12)
growth_signalsSSoT + scoring on the growth CRM — new scoping issue; phases: SSoT → 3-4 highest-yield watchers (ads-runners, revealing-hires, new-DM, site-visitor deanonymization) → if/then plays (Slack alert + sequence trigger + pre-call enrichment) → feedback loop (after #178).- Prompt-level extraction for Atlas growth arms (operator, sales scribe, call center): "Timing + relevance beat personalization. Lead with the dated public event (the signal) that makes the outreach relevant NOW; never lead with flattery. Optimize PCPL, not volume."
- Growth Suite (clinic-facing): the architecture (SSoT → score → enrich → play → loop) is the same shape for clinics' inbound consumer leads: form fill +15, price-page +8, IG engagement +2, no-show risk −X → plays = nurture SMS, win-back offer, front-desk alert. The Zeely ads track generates the demand; this architecture converts it. One design, two tenants.
Related
docs/claude/agent-fleet.md— fleet map; intake-router pattern this mirrors- #178 attribution foundation (identity stitch) — prerequisite for the feedback loop
- #177 sales scribe/voice — consumes signal-enriched leads
- Master Checklist: "Outbound signal-based motion", "Ads bandit", "Repeatable acquisition engine"