HFHS Growth + Customer Care Plan

Hybrid delivery plan - four workstreams build, five outcome demos prove it

The four workstreams from the committed plan run as parallel swim-lanes (that's how the work is owned). The five outcomes are the checkpoint spine: one demo roughly every week, each a complete story assembled from pieces of multiple lanes. Everything built by Friday Sep 4; QA week lands the Gate 2 demo Sep 11.

D1 · Commitments
Wed Aug 5
D2 · Approval loop
Mon Aug 10
D3 · AI home record
Thu Aug 20
D4 · Sales scoring
Fri Aug 28
D5 · Journey + freeze
Fri Sep 4
QA → Gate 2
Sep 8 – 11
How to read this: lanes are the plan's four workstreams plus a legal lane - parallel work, owned by different people. Bar colors show which outcome demo each piece of work feeds. Three of the five outcomes span two workstreams, which is exactly why demos are cut by outcome, not by workstream: "Workstream 2 is 60% done" shows nothing; "watch a promise get made, missed, and escalated" shows everything.

Swim-lanes - who builds what, and which demo it feeds

D1 Commitments (outcome 4)
D2 Approval loop (outcome 5)
D3 AI ingested home record (outcome 3)
D4 Sales scoring (outcome 1)
D5 Journey (outcome 2)
QA / Gate 2
Hard blocker
Workstream ↓ · Week →
Aug 3–7◆ D1 Wed 8/5
Aug 10–14◆ D2 Mon 8/10
Aug 17–21◆ D3 Thu 8/20
Aug 24–28◆ D4 Fri 8/28
Aug 31–Sep 4◆ D5 Fri 9/4 · freeze
Sep 8–11◆ Gate 2 Fri 9/11
WS1 · Growth Intelligence & Conversion
Jack + Wherk
Scoring model design → historical-outcome validation (green by Aug 20)
Coaching pilot · 1–2 reps
Sequence audit
Scoring + routing live, reasons shown
Coaching → all reps
Address-based assessment + automated follow-up sequences
Gold-set validation · calls
WS2 · First Visits & Home Intelligence
Amanda + Westport HMs + Wherk
⚑ Capture standard signed by Aug 7
Master commitment list · owner, due date, evidence
Approval queue + auto-drafted visit summaries + satisfaction check
Ingestion testing with pilot HMs from Mon 8/3 → guided visit flow (offline, gated) + home plan, real-time loop
Inspection leave-behind report
Gold-set validation · first visits
WS3 · Controls & Auditability
James + Wilmer, Emily
Alerting design locked + escalation chain HM→GM→Emily
Alert infrastructure · severity, aging, dedupe, digests
Outbox · retries, failure visibility, duplicate-contact protection
Vendor SLA measurement + aging quotes on the master list
Failure testing · offline, failed providers, retries, dup prevention
WS4 · Measurement, Adoption & Handoff
Emily, James, Wilmer · Wherk support
Gold-set assembly · calls + first visits
Live pilot scorecard · extends existing dashboards
Onboarding materials · HMs + techs · escalation policy final
E2E testing + Gate 2 demo · per-employee compliance view
Legal
Two-party-consent review
⚑ Consent review · starts Mon Aug 3, one week - blocks coaching + ingestion expansion

Dashed bars are hard blockers with named dates: the capture standard unblocks the guided-flow build (D3), the consent review unblocks coaching expansion (D4). The Wherk dev lane threads through all four workstreams in demo order - that's the sequencing, not extra headcount.

Priority key

P0 Must be true for the demo to pass
P1 Committed within the window
P2 The outcome's pressure valve - trimmed first, never squeezed into the next window

The five demos, in delivery order

D1

Tracked commitments Plan outcome 4

Every customer promise has an owner, due date, evidence, and escalation path.
Aug 3 – 5
3 days · demo Wed Aug 5
The demo

Create a promise from a real Westport visit. Watch it appear on the master list with an owner and a due date. Let one go past due - watch it escalate HM → GM → Emily.

  • P0Master commitment list live - owner, due date, evidence, escalation path; single source for alerts and reports - WS2
  • P0Escalation chain wired: HM → GM → Emily; critical items escalate after one miss - WS3
  • P1Vendor promises tracked on the same list (vendor-as-customer) - WS3
  • P2Aging-up-the-chain rules for routine items

Why first: it's the smallest build and the biggest trust win - and every later demo writes promises into it, so each week the list gets visibly fuller.

D2

Human-approved automation Plan outcome 5

AI handles extraction and drafting; humans approve everything customer-facing.
Aug 6 – 10
3 days · demo Mon Aug 10
The demo

AI drafts three visit summaries. Approve one, edit one, reject one - only the approved and edited versions reach a customer. Kill a provider mid-send and watch the outbox retry without double-contacting anyone.

  • P0Approval queue pattern: draft → approve / edit / reject; nothing customer-facing auto-sends - WS2
  • P0Outbox discipline: retries, failure visibility, named owners, duplicate-contact protection - WS3
  • P1Auto-drafted visit summaries as the first application - WS2
  • P1Post-visit satisfaction check with negative-sentiment escalation (writes to the D1 commitment list)

Why second: it's infrastructure disguised as an outcome. The home plan (D3), coaching drafts (D4), and follow-up sequences (D5) all run through this queue - building it once here means the later demos inherit it.

D3

AI ingested home record rolled out Plan outcome 3

One trusted record per home, built from recordings, notes, photos, and staff judgment.
Aug 3 – 20
testing from Mon Aug 3 · build Aug 11–20 · demo Thu Aug 20
The demo

An HM completes a guided Westport visit - checklist, photos, notes, offline, no completion until requirements are met. In real-time, the AI-generated home plan (profile updates, repairs, projects, promises) lands in the approval queue; approve it and watch the home record and commitment list update.

  • P0Guided mobile visit flow - capture standard enforced, works offline, completion gated - WS2
  • P0Ingestion engine: recordings + notes + photos → proposed profile updates, repairs, projects, promises - WS2
  • P0Real-time turnaround loop running with pilot HMs through the D2 approval queue - beats the plan's stated one-business-day commitment
  • P1Customer-facing and internal views of the record - WS1 standardized profile
  • P2Home health / QBR generation - sequenced last by the plan's own rule; the designated cut if this window runs hot

Why the longest window: this is the ten-day outcome. Ingestion testing with the Westport HMs starts Monday Aug 3 and runs underneath D1 and D2, so by the time the dedicated build window opens Aug 11 the extraction quality is already field-tested - the build week is the guided flow, the real-time loop, and the record views. The dedicated build still only starts on time if the capture standard is signed by Aug 7 (dashed flag in the WS2 lane).

D4

Sales decision support Plan outcome 1

Every lead gets a transparent recommendation: fit, relationship potential, readiness, confidence, membership level.
Aug 21 – 28
6 days · demo Fri Aug 28
The demo

Open a live lead: score, reasons, and recommended membership level on screen. Show the validation numbers against historical outcomes. Watch routing send it to the right rep because of the score, not despite it.

  • P0Scoring live with reasons shown (home value, age, member proximity) - WS1
  • P0Routing driven by score - only after validation passes (the plan's own gate) - WS1
  • P1Membership-level and conversion-readiness recommendation on every lead
  • P1Coaching expanded to all reps - requires the consent review closed (legal lane, week 1)

Why six days is enough: Jack's lane runs the scoring design and historical validation during weeks 1–3 (the long purple bar in WS1), so this window is build-and-wire, not research. If validation isn't green by Aug 20, this demo ships "reasons shown, routing manual."

D5

Unified conversion journey Plan outcome 2

Address → assessment → inspection → report → follow-up → membership runs as one pipeline.
Aug 31 – Sep 4
5 days · demo Fri Sep 4 = code freeze
The demo - this is the Gate 2 rehearsal

Type an address → instant house profile → lead scored and routed → free inspection booked → guided visit → leave-behind report → automated follow-up drafted and approved → membership close. Every piece was demoed alone in D1–D4; this demo proves they're one pipeline.

  • P0Address-based assessment as the conversion event - instant profile, free inspection as the standard moment - WS1
  • P0Pipeline wiring end to end - each stage hands off to the next with no manual re-entry - WS1 + WS2
  • P1Automated follow-up sequences on the mapped journey (through the approval queue)
  • P1Inspection leave-behind report - WS2

Why last: it's the integrating outcome - it can only be true once the other four are. Five days is enough because it's wiring demoed pieces together, not building new ones. Friday's demo doubles as the freeze: whatever passes is what QA hardens.

Workstream → outcome coverage

All five outcomes are fully covered by the four workstreams - but three of the five span two workstreams, and WS4 produces no outcome of its own: it proves the other four (gold sets, E2E testing, scorecard, onboarding). That's why the build is organized by workstream and the checkpoints by outcome.

Plan outcomeBuilt byDemo
1 - Sales decision supportWS1 (scoring, routing, coaching)D4 · Aug 28
2 - Unified conversion journeyWS1 (assessment, follow-up) + WS2 (first visit)D5 · Sep 4
3 - Single verified home recordWS2 (guided flow, home plan) + WS1 (standardized profile)D3 · Aug 20
4 - Tracked commitmentsWS2 (master list) + WS3 (vendor promises, alerts, escalation)D1 · Aug 5
5 - Human-approved automationWS2 (staff approval) + WS3 (outbox, duplicate prevention)D2 · Aug 10
Verification of all fiveWS4 (gold sets, E2E testing, scorecard, onboarding)Gate 2 · Sep 11

QA week - Sep 8 to 11

No new build. Gold-set validation (calls + first visits, scored for accuracy, omissions, hallucination), end-to-end failure testing (offline visits, failed providers, retries, approvals, negative feedback, duplicate prevention), pilot scorecard live, onboarding materials out. Friday Sep 11: the Gate 2 demo - the same run as D5, on a hardened system. Contractual checkpoint Mon Sep 21.