The Gameplan · built 2026-08-11 · runs Aug 17 → Nov 13, 2026 · pays through Feb 2027
Three months of building. Six months to double.
Baseline (to be confirmed at the Aug 13 discovery day): ~$55k/month, ~300 clients, 5 people. Target: $110k/month by February 2027. The strategy: build the machine in Month 1, amplify in Month 2, convert in Month 3 — so AEO visibility, outreach, referrals and nurtured leads all peak together while renewals hold the base.
Sources: approved team-meeting strategy (2026-08-06) · operating chart ($63k/$85k/$120k, 2026–28) · industry deep-dive & SWOT · ABA coverage research · 3-month revenue framework (Aug 2026) · NotebookLM research notebook.
1 · The number, month by month
Doubling is not a February event — it compounds. Slow at first while the machine gets built, steep at the end when conversion, referrals and content visibility stack. The operating chart's 2026 goal ($63k) falls in October; we pass it on the way.
MRR trajectory — $55k → $110k
Monthly targets. Solid = this plan; dashed line = the 2026 operating-chart goal.
Where the +$55k comes from — the revenue bridge
A model, not a promise — the Aug 13 discovery day replaces these splits with real numbers. Ranked by cost of acquisition: selling more to the 300 clients we already have comes first.
2 · The client lifecycle — and what serves each stage
Every dollar of the doubling flows through six stages. Under each stage: the systems that serve it — solid border means live today, dashed means this plan builds it.
3 · Client value — why cross-sell is lever #1
Illustrative 3-year value model (discovery day replaces with actuals; 94% 3-yr retention from the operating record). The same client is worth roughly 2.6× with a second pillar attached — and acquisition cost for a cross-sell is a phone call, not a campaign. 15 cross-sells ≈ a third of the entire target.
3-year value per client, by pillars held
The pipeline math that has to hold, Month 3 onward
4 · The 3-month build — who builds what, when
Month 1 Foundation & Systems · Month 2 Launch & Amplify · Month 3 Optimize & Convert. Gate reviews at each month-end: the listed outcomes are the ticket to the next phase.
Workstreams (Aug 17 → Nov 13)
| Month | Infrastructure (Pulseaa builds) | Go-to-market (team runs) | Gate — what must be true |
|---|---|---|---|
| M1 Foundation | Discovery-day data in · CRM verified & exported · renewal calendar trued up · lead-alert Cloud Function · scoreboard dashboard v1 · onboarding checklists · ICP defined from top clients · AEO baseline scored · cross-sell list of 60 names | LinkedIn profiles optimized · first cold sequences (Apollo) · networking calendar mapped (Chamber, investor groups) · same-day-call SLA in force · Substack claimed and first sends | 5–10 qualified conversations · dashboard reviewed in first Monday meeting · every lead answered same business day |
| M2 Amplify | Nurture sequences live · ABA prior-auth tracker · renewal watchdog agent · CMS Marketplace key live in /quote engine · webinar/event landing page · Spanish sign-off shipped | Content 2×/wk under principals' bylines · scale outreach to 100–150 contacts/wk · referral program launched · one webinar or local event hosted · retarget magnet leads | 15–25 qualified conversations · 3–5 proposals out · 2 local partnerships · both funnels converting measurably |
| M3 Convert | A/B message testing · AI follow-up automation · reporting agent (weekly scorecard email) · commission reconciliation check · case-study pages | Close first cohort (20–30% of pipeline) · collect testimonials with written consent · double down on best channel · referral flywheel running | 3–5 deals closed · MRR ≥ $70k · case studies feeding AEO · pipeline self-sustaining into Dec–Feb compounding |
5 · The content & authority engine
One 3–5 minute voice note becomes an article, a newsletter, clips, and AI-answer visibility. The principals already hold the authority assets most competitors can't fake — two books, 30 years, real ABA advocacy stories.
The flywheel
Already in the arsenal: 5 principal-bylined resource articles · the bilingual Texas ABA Coverage Playbook lead magnet · the /aba screener and /aca-review funnels · FAQ schema + llms.txt AEO plumbing on every page. M1 adds the 100-question FAQ knowledge base (feeds both assistants AND AI-search answers); M2 turns the cadence on; M3 feeds closed-deal case studies back in — with written consent, per the standing FINRA rule: never a fabricated quote or figure.
6 · The AI agent roster
Rule of the house: agents educate and automate — a licensed human makes every recommendation, and no agent ever speaks as Michael or Justin.
ACA Assistant
Voice+text educator on 14 pages. 2026 penalty figures, filing rules, hands off to the free ACA review. Now says "review", never "audit".
ABA Coverage Assistant
Chat-only, bilingual, Michael's photo with hard "I am an AI" rules. Medicaid/CHIP/MBIC, OOP-max reality, mandate truth. Hands off to /quote.
ABA Match Engine
Plan-match + cost model behind /quote and /aba: 38 unit tests, CMS Marketplace adapter, advisor-reviewed output only.
Lead Sentinel
Cloud Function: every new Firestore lead → instant email/SMS to the team + same-day-call SLA timer on the dashboard. The highest-ROI 50 lines of code in this plan.
Outreach Copilot
Apollo-driven ICP lists (from the real top clients), AI-personalized sequences, omnichannel cadence: email → LinkedIn → call. Human sends, agent drafts.
Renewal Watchdog
Reads the verified renewal calendar; opens a ticket + prep checklist at 90/60/30 days; flags any renewal nobody has touched. Retention is half the doubling.
Nurture Sequencer
Playbook and magnet leads get bilingual drip sequences tied to their situation (ABA family vs employer), ending in a booked call, not a pitch.
Scorecard Reporter
Monday 7am: last week's leads, conversations, proposals, closes, MRR vs the trajectory line — mailed to Michael, Justin and MT. No meeting prep, no hiding.
Marketplace Live Data
The /quote engine runs on sample data until the free CMS Marketplace API key lands in Secret Manager. Requested 2026-08-13 — CMS issues it by email, so this is now a waiting game, not a task.
7 · System integration — one nervous system
Everything lands in one spine: source → Firestore → back office → CRM → carriers. Solid = live; dashed = built in the month marked. The Aug 13 discovery day answered the one unknown that mattered — see §10–§13 for what we are building into that spine.
8 · Staying on track — the scoreboard and the ritual
A plan without a weekly number is a wish. Eight numbers, one owner each, reviewed every Monday against the trajectory in section 1 — in the back office, where the revenue log and targets already live.
| # | Metric | Baseline (8/13) | Feb '27 target | Owner | Source |
|---|---|---|---|---|---|
| 1 | MRR ($k) | ~55 — confirm | 110 | Michael H | /admin revenue log |
| 2 | New leads / week | confirm 8/13 | 25+ | MT | Firestore leads |
| 3 | Same-day response rate | unknown | 100% | Team | Lead Sentinel SLA |
| 4 | Qualified conversations / mo | confirm | 20–25 | Michael H + Justin | CRM |
| 5 | Proposals out / mo | confirm | 8–10 | Michael H | CRM |
| 6 | Closes / mo · win rate | confirm | 3–5 · 20–30% | Michael H | CRM |
| 7 | Clients with 2+ pillars | confirm (the gold number) | +15 | Justin | CRM |
| 8 | Renewals saved / at-risk touched | — | 100% touched ≥60d out | Team | Renewal Watchdog |
| 9 | Google reviews (count · rating) | 2 · one negative | 50+ · 4.7★+ | Whole team (§9) | Google Business Profile |
The cadence
| When | What happens |
|---|---|
| Monday, 15 min | Scorecard read against the curve. Every red number gets one action and one owner before the meeting ends. (M3 onward the Scorecard Reporter agent mails it at 7am.) |
| Month-end gates | Gate review against section 4. Passing the gate is the permission to spend on the next phase; missing it triggers a re-plan, not a shrug. |
| Every release | Antigravity runbooks, checks green, Notion hub updated. Compliance rule standing: nothing public that Ameritas hasn't seen where required. |
Risks we're naming out loud
| Risk | Mitigation |
|---|---|
| Capacity — 5 people at 2× volume | Onboarding checklists + ticket queue + agents take the repetitive load first (M1); hiring decision at Gate 2 with real numbers. |
| Renewal-season crunch eats the outreach | Renewal Watchdog starts prep at 90 days; outreach volume flexes down, never to zero, in heavy months. |
| Compliance (FINRA/Ameritas) | Standing rules on the Notion hub; nothing fabricated ever; case studies only with written consent; disclosure review is an open item with a date. |
| Key dependencies | CMS API key (requested 8/13, awaiting CMS email) + Health Sherpa ONE key (free, self-serve) · BenefitZone credentials for the extraction · Substack claim · Spanish native-speaker sign-off · Ameritas written OK on the review programme (§9). All on the open-items list with owners. |
| Subsidy-cliff market shock (2026) | Also an opportunity: families and employers need advisors more this year than any year since 2014 — the ABA and ACA funnels are pointed exactly at that pain. |
9 · The reputation engine — from 2 reviews to 50
Today Google shows 2 reviews, one of them negative — to a stranger, a coin flip. Industry reality: agencies at 50+ reviews and 4.5★+ get ~3× more inbound from Maps, and 71% of consumers won't call an insurance agency under 4 stars. Every referral checks Google before dialing, so this multiplies everything in sections 2–5. Google and Yelp have opposite rules, and the securities registration adds a third — all three are built into the plays below.
Review trajectory — steady drip, never a burst
~2–3 asks a week at the four moments below, at a ~30% say-yes rate ≈ 6–8 new reviews/month. A sudden spike on a 2-review profile reads as purchased — to Google's filters and to humans. Velocity is the strategy.
| The four asking moments | Who asks | The ask (personal text or email, within 2–4 hours — never a blast) |
|---|---|---|
| ABA appeal won / therapy authorized | Michael or Justin | EN: "So glad we got this sorted for [child's first name]. If our work helped, a Google review helps the next family find us — it takes one minute: [link]" · ES: "Nos alegra haberlo resuelto. Si nuestro trabajo le ayudó, una reseña en Google ayuda a la siguiente familia a encontrarnos — toma un minuto: [link]" |
| Enrollment / onboarding wrapped | Account manager who ran it | "That's everything set up and live. If the process was smooth, would you share that in a quick Google review? [link]" |
| Renewal saved / rate fight won | Whoever fought it | "Happy we could keep the increase off your back this year. A quick review helps other Houston employers find us: [link]" |
| ACA penalty avoided / filing cleaned up | Connor / filing lead | "Your filings are in and clean. If that took a weight off, a one-minute Google review means a lot to a small practice: [link]" |
Compliance rails — these are rules, not suggestions
| Rule | Why / how |
|---|---|
| Yelp: never ask. Anyone. Ever. | Yelp prohibits all review solicitation — verbal, email, signage, tools — and enforces with ranking penalties and public consumer alerts. Yelp play: claim the page, complete it, respond professionally, organic only. |
| Google: no gating, no incentives | Ask everyone at the chosen moments — the filter is the moment, never a "were you happy?" pre-screen (gating violates Google policy and FTC rules). Nothing of value in exchange, ever — no gift cards, no discounts, no raffle entries. |
| Scope: insurance side only | Asks reference benefits, ACA, payroll, enrollment, ABA advocacy. Never prompt anything about 401(k), investments, or advisory performance — testimonials touching securities trigger SEC marketing-rule and FINRA 2210 obligations. |
| Ameritas sign-off before launch | One-page program description to Ameritas compliance (moments, scripts, scope exclusion) — get written OK. Add to Thursday's Block 6 and the open-items list. The programme does not start until this is back. |
| Responses name no client details | Michael responds to 100% of reviews within 48h. Negative: calm, no case details ("Please call me directly — Michael, 281.493.6862"), never confirm someone is a client. Insurance work is confidential even when the reviewer isn't. |
| Never fabricate, never import | The standing rule for this client applies here with force: no invented reviews, no staff/family reviews, no buying. One fake review discovered costs more than fifty real ones earn. |
| When | Rollout |
|---|---|
| M1 · wk 1 | Verify Google Business Profile ownership (discovery-day access register) · complete profile (four pillars as services, ES description, photos) · get the g.page review short-link · flag-check the negative review against policy; if legitimate, Michael posts the calm response · claim the Yelp page, respond, and never touch it again. |
| M1 · wk 2 | Ameritas one-pager submitted · bilingual QR review card printed for the office and enrollment packets · scripts above loaded as saved snippets in everyone's phone/email. |
| M1 wk 3 → | Programme live: 2–3 asks/week at the four moments · reviews count joins the Monday scoreboard (KPI #9) · every new review answered within 48h. |
| M2–M3 | Best review lines become AEO fodder (with the reviewer's OK) · review count added to proposal decks · Scorecard Reporter agent tracks count + rating weekly. |
10 · The client login — what a client can actually do
From the 8/13 call: “train clients to log in to do everything, because then there’s always a log.” That is the whole thesis. A phone call is not a record; a portal thread is. Rebuilt 2026-08-13 as a real account: you sign in, we resolve which client you are, and every panel renders your own record from Firestore. No sample data anywhere. Staff grant access in /admin → Clients; the contract is docs/PORTAL-DATA-MODEL.md and 21 permission-boundary tests guard it.
| Capability | What it replaces | Status | Who asked for it |
|---|---|---|---|
| Secure message thread | Phone calls nobody logged; email buried in one person’s inbox | Built — append-only, both directions; staff reply from /admin | Justin — “there’s always a log” |
| Member changes | The memberchanges@hipson.com inbox as an unmanaged queue | Built — portal form writes a tracked ticket; email path still works and becomes a ticket once the mail hook ships | Justin + Jessica |
| Coverage & policy view | “What plan am I on?” calls; screenshots of ID cards | Built — renders the client's real policies; staff maintain them in /admin | Michael H. |
| Documents & e-signature | Consent forms chased by email; Medicare consents stored ad hoc | Surface built · signing engine M2 · consent PDFs owed by Justin | Jessica — “anything to sign for CYA?” |
| Secure file upload (census) | Census spreadsheets with SSNs emailed in plain text | Surface built, deliberately disabled — needs Firebase Storage enablement, its own reviewed change | Michael H. (PII risk) |
| Submission / application status | “Where’s my application?” calls during enrollment | Requests render live with status; application status arrives with the Health Sherpa webhook (§13) | Jessica |
| Self-serve help | 40-minute hand-holding calls through a marketplace form | Live — /personal/enrollment, glossary, carrier directory, 5 languages | Jessica — “they’re helpless on the phone” |
The rule we designed to
Optional, never mandatory. Justin: “I don’t want to force clients to log in — I’m afraid they won’t like it — but I’d like them to have the option, and I don’t want to create and troubleshoot logins for every single person.”
11 · The back office — what the team can actually do
Live at /admin behind Google or password sign-in (the team is on Microsoft 365, so password accounts are created by hand until SSO). Every tab below exists today unless marked. This is also the shell the CRM grows inside — see §12.
| Tab | What it does | Status |
|---|---|---|
| Overview | KPI row, needs-attention list, latest leads, the Monday ritual, shortcuts | Live |
| Gameplan | Revenue curve vs the baked trajectory, click-to-edit scorecard, 24 milestone gates | Live |
| Leads | Every funnel lead — filter, status, notes, CSV, drawer. Auto-research (company + public profile pulled on arrival) is M1 | Live · enrichment M1 |
| ABA quotes | Family intake → licensed advisor runs the plan-by-plan match | Live on sample data · real plans when the CMS key lands |
| Member changes | Adds, terminations, address changes — open/working/done, confirmation in writing | Live (new) |
| Clients | Create clients, grant/revoke portal access, reply in the thread, move request statuses, publish documents — every action audited | Live (new) · this IS the CRM seed, §12 |
| Brand vote | Logo vote + comments from /brandguide, deduped one-per-person | Live (new) |
| Tickets | Site review comments from the ?review=on widget | Live |
| Targets · Growth plan · Assets · Agreement | Revenue log, 3-month plan completion, funnel/flyer links, signed agreement record | Live |
| Renewal calendar | Every group’s renewal date with a 120/90/60/30-day watchdog — the anti-autopilot machine | M1 |
| Commissions | Import carrier statements, reconcile against the book, flag missing commissions | M2 — see §12 |
| Plan ranking | Upload the year’s approved benefit summaries → ranked “our top plans this year” list, client-facing tiers | M2 — Justin’s open-enrollment ask |
Access, honestly stated
Rules are server-enforced and name the four team addresses directly, OR’d with an allowlist doc so people can be added without a deploy. Two gaps remain, both scheduled: Microsoft 365 SSO (the team’s mail is not Google, so passwords are hand-created in the console today) and role separation — Jessica completes the discovery worksheet but must not see the money block. Neither is a code problem; both need a decision and ten minutes in a console.
12 · The Hipson CRM — build, don’t rent
Two things were said on 8/13. Mike: “I want to switch us completely to AgencyBloc as our sole CRM” — the blocker being that BenefitZone, what they run today, does not let them export. Then Justin, looking at the back office: “am I crazy to think you’re pretty close to just building an entire CRM? Do we need to be paying for another one?” That is the decision this section prices out.
| AgencyBloc AMS+ capability | What it means for Hipson | Our build |
|---|---|---|
| Contacts, leads, clients, policies with activity history and custom fields | The book of business — individuals, groups, dependents, one timeline per relationship | M1 — `clients` collection exists; extend to policies + activity |
| Group and individual policy administration, carrier tracking, lifecycle | Both sides of Hipson in one system — group medical and the ABA/individual families | M1–M2 |
| Commission processing — import carrier statements, map splits, spot missed commissions | The single highest-dollar feature. Carriers underpay and nobody catches it without reconciliation | M2 — statement import + reconcile vs the book |
| Quoting that syncs across the platform | Already ahead here: /quote, /aca-review, /aba and the CMS + Health Sherpa engines are ours | Live |
| Workflow automation and task assignment | Renewal watchdog, member-change SLAs, onboarding checklists | M1–M2 |
| Open API, read/write clients, policies, commissions, tasks | Ours is the database — no API needed to reach our own data, and no export permission to ask for | Inherent |
| HIPAA-standard storage, access controls, audit trail | Non-negotiable — child health data lives here | Rules + audit log M1; a formal review before real PHI lands |
The money, stated plainly
AgencyBloc AMS+ is published from about $109 per user per month at entry (older public tiers listed $59 and $79); real pricing is quoted per agency. At five users that is roughly $5,400–$6,500 a year, forever, rising with headcount — and the book still lives on someone else’s server under their export terms. Building it is a one-time effort inside a system we already run, with no per-seat tax as the team grows. Verify the live quote before this is treated as a decision — published entry pricing is not a proposal.
| Phase | What ships | Proves |
|---|---|---|
| Phase 0 now | `clients` records + portal access keyed to email; member changes and support tickets already flowing into the same database | The shell is real, not a mockup |
| Phase 1 M1 | BenefitZone extraction agent — signs in with their credentials, pulls every group, policy and contact BenefitZone won’t export, lands it as `clients` + `policies`. Renewal calendar built from it. | The book is ours again — the one thing blocking any move |
| Phase 2 M2 | Commission import + reconciliation · policy lifecycle · activity timeline · document store · task/SLA automation | Parity on the features that actually get used daily |
| Phase 3 M3 | Full switch: the team works from ours, BenefitZone is read-only, then cancelled. Reporting agent emails the weekly scorecard. | One system, no per-seat bill, no export permission needed |
Honest risk: AgencyBloc is a mature product with years of edge cases in it — carrier-statement formats alone are a long tail. The mitigation is sequencing: extract first (that is pure gain regardless of the decision), then build the features they actually touch, and keep AgencyBloc as the fallback if commissions prove deeper than a month. If the team wants to buy instead, the extraction work is not wasted — it is exactly what the migration needs.
13 · The data layer — CMS, Health Sherpa & the carriers
Two free APIs turn the quoting tools from “illustrative” into “accurate”, and feed the CRM without anyone retyping. Both are server-side only: their keys never touch a browser.
| Source | Cost | What we get | Status |
|---|---|---|---|
| CMS Marketplace API developer.cms.gov | Free · self-serve key | Every 2026 individual plan by county, premiums, cost-sharing, plus provider and drug coverage lookups — what /quote needs to stop running on sample data | Proxy built (`functions/index.js`, admin-gated, allowlisted paths) · key REQUESTED 2026-08-13 — CMS sends it by email; then one secret + one functions deploy |
| Health Sherpa ONE one.healthsherpa.com | Free for ACA quoting · enrollment is approval-gated | /v1/quotes, county and issuer reference, and — once approved — a hosted, pre-filled enrollment session we hand the family into, plus webhooks for application and policy status | Confirmed available (8/13). Key is self-serve; build M1, enrollment application M2 |
| Carrier statements | Existing rails | Commission reconciliation against the book (§12 Phase 2) | M2 — needs a sample statement from each carrier |
| Apollo.io | Paid tier TBD | B2B enrichment: HR/CFO contacts by company size and region; enriches the 5,500-filing list Mike is sending | M1 pilot on their real list |
How a family flows through it
| Before we design around enrollment | Why it matters |
|---|---|
| Written confirmation of production access, rate limits, and whether branded/embedded enrollment is allowed | Quoting is self-serve; enrollment is approval-gated and shapes the whole funnel design |
| FFM ID + National Producer Number | Required to process applications and reach post-enrollment functions — Justin/Michael H. supply these |
| Lead ownership and data-export terms in writing | We just spent a phase escaping exactly this problem with BenefitZone. Do not re-enter it blind. |
| Webhook endpoint, auth method, event subscriptions | This is what makes portal application status live rather than a phone call |
| HIPAA and consent posture reviewed for applicant data | The individual/families consent form already exists — it now has to cover API-collected data too |
Sequencing recommendation: ship the free quote API first and validate conversion on our own funnel, then apply for the hosted enrollment session rather than fully programmatic enrollment — smaller build, cleaner compliance boundary, and it still keeps the acquisition funnel ours.