The 114 lessons before this one taught the discipline. This lesson is the artifact. A dated, named, metric-tied 90-day plan the operator walks into the next board meeting or PE-partner review with — covering the platform AI rollout, the governance docs, the org-chart changes, the data-center positioning thesis, and the first five KPI lifts the operator commits to deliver. This is the capstone of the AI for Skilled Trades & Home Services program. The operator finishing this lesson does not write a memo about AI; they walk out with a 90-day operating plan their executive team, PE partner, lenders, and franchise HQ can audit weekly. The plan synthesizes every named workflow from L1 through L5: missed-call recovery (Avoca / Jobber AI Receptionist / Housecall Pro AI Agents / ServiceTitan Voice), dispatch optimization (Dispatch Pro / Sera profit-aware scheduling), ride-along coaching (Rilla), RC&D triage with SLA tiers, marketing attribution (CallRail + Ryze AI + Hatch + NiceJob + Podium + Birdeye), the financing tier matrix (Wisetack / GreenSky / Synchrony), the pricebook auto-update workflow, the fleet 179 plan, the M&A DD-to-Day-100 workflow, the hyperscale bid-prep workflow, the cybersecurity breach response playbook, and the four-stage rollout framework. Every claim ties to a row in a one-page milestone table that survives a Wrench Group, Authority Brands, Apex Service Partners, Sila Services, Path Light Pro, Redwood Services, Leap Partners, or ARS-Rescue Rooter quarterly business review. Day 0 is when the operator signs the plan. Day 90 is when the operator defends the metric movement against it. The plan is the contract.

Why 90 Days and Not 12 Months

The L4 roadmap covers 12 months across four 90-day phases because that is the cadence at which a single shop builds AI competency from baseline. The L5 capstone plan is a single 90-day window for a different reason: the operator running 5, 25, 100, or 450 locations does not commit to a 12-month sequence at the platform level — the operator commits to the next 90 days with explicit milestone gates, then commits to the next 90 days against the movement the first 90 produced. The platform plan is built quarter by quarter against the PE board's cadence, not against the shop's annual budgeting cycle.

90 days is also the unit at which AI deployment produces defensible metric movement at platform scale. A pilot location runs 60-90 days. A reference location runs 60 days. Wave 1 runs 90 days. The four-stage rollout framework converges on 90-day decision cycles because that is the cadence at which the platform distinguishes signal from noise, proceed from hold from roll-back, and operator readiness from operator drift. A 12-month L5 plan dilutes the discipline; a 30-day plan does not produce enough cycles to measure.

The 90 days also align to the operator's external reporting obligations. PE-backed platforms report quarterly. Franchise-mandated operators report quarterly to franchisor HQ (Authority Brands QBR, Wrench Group portfolio review, Apex synergy synthesis). Lenders covenant test quarterly. The plan is slide 4 on the board deck, the document the PE partner reads before the call, the document the franchise field consultant references at the QBR. The cadence is the heartbeat of external accountability.

The Day Zero Deliverables

Day 0 is the day the operator signs the 90-day plan. Six deliverables must exist on Day 0 — not in draft, not in concept, but in production form, signed by named owners. Without Day 0 deliverables, the plan is aspiration; with them, the plan is operational.

Deliverable 1 is the executive summary. One page, four paragraphs: the 90-day thesis (which workflow deploys next, which KPI it lifts, against which baseline); the five KPIs and target delta for each; named owners per workflow with reporting-line clarity; rollback triggers and named decision authority. The PE board reads this first.

Deliverable 2 is the location-by-location pilot map. A grid with rows for each location and columns for the workflow deployed (Avoca, Rilla, Dispatch Pro, Hatch), the readiness tier (A/B/C from the four-stage framework's spread analysis), the named deployment owner, the first-metric-movement milestone date, and the rollback authority. Every PE board asks "which locations and when?" The grid answers it.

Deliverable 3 is the governance and policy docs. The platform AI policy signed by the operator, CISO, CFO, and PE operating partner. The vendor governance matrix: every AI tool with master agreement status, SOC 2 review date, sub-processor map, MFA enforcement, data retention policy, breach trigger. TCPA / Reg Z / FCRA compliance attestation per state of operation. The cybersecurity breach response playbook signed and tabletop-tested within the prior 6 months.

Deliverable 4 is the org-design recommendations. Director of AI Operations with reporting line, comp band, and named hire (or named search). Conversation QA Lead, Prompt Librarian, and Data Center Account Director with reporting line, comp band, named hire or hiring plan. The redesigned CSR / dispatch / tech team org chart: centralized AI inbound + local CSR escalation, centralized scorecard QA + local coaching, shared prompt libraries with local overlays. The workflow is not just a vendor contract; it is a redesigned operating model.

Deliverable 5 is the data center positioning thesis. A two-page memo answering: (1) does the platform have electrical capacity — IBEW or open-shop labor, prevailing-wage exposure, prime-vs-sub posture — to bid hyperscale? (2) Which named hyperscalers (Microsoft, AWS, Meta, Google, CoreWeave, QTS, Equinix) or named primes (Rosendin, Cupertino Electric, MMR) are realistic 12-24 month bid targets? (3) What capital and operating commitments does entry require? (4) What is the alternative-path EBITDA for staying in residential / light-commercial? (5) What is the decision and timeline?

Deliverable 6 is the five-metric board dashboard. Five KPIs, one page, one chart per KPI showing baseline, target, current, trend: platform-weighted booking percent, MPR, financing close on $5K+ jobs, recall percent, GLSA ROAS or RPL by source. The five chosen on Day 0 are the five the board sees every month; switching mid-quarter is the discipline failure PE boards penalize hardest.

Day 30 — The Deployment Checkpoint

Day 30 is the first stage gate. The operator and executive team review three questions in a 90-minute meeting: did the Day 0 deliverables execute, are the deployment milestones on track, and what changes does Day 31-60 require based on what the first 30 days surfaced. The Day 30 review is the operator's first opportunity to course-correct without losing the quarter.

Day 30 metric movement is small but measurable. Avoca deployment: 5-10 sites in CSR training and integration by Day 14, Avoca live at 60-80% of wave 1 sites by Day 30, first booking-percent and missed-call-percent baselines captured. Rilla deployment: 5-10 advisor teams trained on recording discipline by Day 14, first 30-40 virtual ride-alongs per manager logged by Day 30. Dispatch Pro deployment: skill tags, revenue history, and capacity rules configured by Day 14, Dispatch Pro re-evaluating boards every 10 minutes by Day 21, first RPT baseline by Day 30. The checkpoint is about deployment milestone discipline, not full lift visibility.

The review surfaces deployment exceptions. Which wave 1 sites missed Day 14 training and why? Which vendor success managers are not engaging at the contracted cadence? Which named owners are stretched across colliding deployments? The exception log is the operator's first window into operating-bench absorption — deploying faster than the bench absorbs vs. under-utilizing the runway. It becomes the input to Day 31-60 corrective actions.

The Day 30 board update is one slide. Wave 1 sites deployed against plan, exceptions logged, corrective actions taken, the five KPIs at first-baseline, Day 60 milestone commitments. The board hears progress, not lift — lift comes at Day 60 and Day 90. A Day 30 update claiming lift is a flag the operator does not understand the deployment cadence.

Day 60 — The Mid-Cycle Lift Measurement

Day 60 is when deployment lift becomes measurable. The operator and executive team meet for two hours. Three questions: which workflows are producing the lift the thesis required, which are off-trend with what corrective action, and what is the Day 90 forecast against board KPI commitments.

Day 60 metric movement is meaningful. Avoca sites typically show missed-call percent at 8-12% (against 22% baseline), after-hours capture at 50-65%, booking percent up 6-10 points. Rilla advisor teams show close-rate movement of 4-8 points (against the 8-18 point full lift at Day 90). Dispatch Pro deployments show RPT movement of 6-10%. The Day 60 lift is not the full Day 90 lift, but it is enough to distinguish "on trajectory" from "off trajectory" — and that distinction drives Day 61-90 corrective actions.

Day 60 spread analysis is the first look at site-level distribution. Wave 1 deployed 8 sites; Avoca lift ranges from 14 points booking movement at the top to 3 at the bottom. The 11-point spread is the diagnostic. Four explanatory variables (CSR-floor maturity, local market difference, GM sponsorship, tech-stack version drift) get assessed against bottom-quartile sites. The spread informs whether the platform commits to wave 2 segmentation at Day 90 or whether wave 1 needs another 30 days.

The Day 60 board update is two or three slides. Wave 1 lift by site (top vs. bottom quartile), spread analysis, exception handling against Day 30 commitments, Day 90 KPI forecast, the wave 2 readiness tier hypothesis. The board hears "tracking against thesis; here is the spread; here is wave 2; here are the risks." A Day 60 update without spread analysis is a flag the operator is reporting average lift and missing the deployment quality story.

Day 90 — The Board Defense

Day 90 is the operator's quarterly board defense. Four hours with the executive team to prepare; two hours with the board to defend; the rest of the quarter to execute against the next 90-day commitment. The Day 90 review is the operator's accountability moment, and the deliverables are five.

Deliverable 1: five KPI movement against commitment. Booking percent: target X, actual Y, delta Z, here is why. Same structure for MPR, financing close, recall, GLSA ROAS. The board's first slide.

Deliverable 2: wave 1 site-level lift documentation. The 5-10 sites deployed, workflow-specific primary metric per site (Avoca missed-call %, Rilla close rate, Dispatch Pro RPT, Hatch reactivation), aggregate margin contribution, deployment cost against lift, spread analysis, and the wave 2 readiness tier plan with sites segmented A/B/C. The slide that converts AI from expense to EBITDA line item.

Deliverable 3: governance attestation update. Master agreements signed against Day 0 commitment, SOC 2 reviews completed, MFA audited, sub-processor map current, breach response tabletop completed. The slide that converts AI from risk to managed-risk.

Deliverable 4: org-design execution update. Director of AI Operations hired (or named search at term), Conversation QA Lead, Prompt Librarian, and Data Center Account Director hired (or alternative-path thesis defended), the redesigned CSR / dispatch / tech team org chart deployed at wave 1 sites with first-cycle operating metrics. The slide that converts AI from project to platform competency.

Deliverable 5: the next 90-day plan. The Day 91 commitment. Which workflow gets piloted next (Hatch if Avoca was first, Dispatch Pro if Rilla was first), which sites move from wave 1 to wave 2, which segmentation tier each wave 2 site occupies, which named owners run the next 90 days, what the five KPIs commit to at Day 180. The slide that earns the next 90-day investment runway.

The Day 90 defense is the moment the operator compounds credibility or loses it. Operators who defend with spread analysis and documented dependencies that did not close earn the board's trust for the next 90 days. Operators who report average lift without analysis lose it quickly. Compounding credibility across four quarterly defenses distinguishes operators who become long-tenured platform CEOs from those replaced at the 18-month mark.

The Seven-Question Gate Before Each Milestone

Between Day 0, Day 30, Day 60, and Day 90, the operator runs every wave 1 site, every workflow deployment, and every governance decision through a seven-question gate. The gate prevents deployment drift between checkpoints. Skipping the gate is the failure mode this capstone protects against.

Question 1: does this milestone move one of the five board KPIs, or is it operational hygiene? If not a board KPI mover, the milestone gets deferred or delegated. The operator's calendar capacity is the platform's scarcest resource.

Question 2: is there a named owner with reporting-line accountability and a documented metric target? No named owner means no accountability. No metric target means unmeasurable.

Question 3: has the dependency chain been verified? Dispatch Pro depends on clean CSR booking from Avoca; Ryze AI depends on clean closed-revenue from CallRail + ServiceTitan. Deployment on broken substrate produces ambiguous lift.

Question 4: what is the rollback trigger, who decides, and what is the customer-facing language if rollback executes? Every milestone has a rollback path; not every milestone has a documented one. The documentation is the discipline.

Question 5: does the deployment vendor's success manager commit to the milestone date in writing with named contact and escalation path? Vendors who will not commit in writing are not aligned with the platform's stage-gate discipline.

Question 6: what is the change-management absorption load on the operating bench in the next 30 days? Tool-sprawl management is the wave 2 metric that emerges precisely because operators under-estimate operating-bench absorption.

Question 7: how does this milestone update the five-metric board dashboard? Milestones that do not update the dashboard are invisible to the platform's external accountability layer.

The gate runs at every checkpoint and every site-level deployment kickoff. Operators who run it consistently compound discipline across the quarter; operators who skip it accumulate exceptions that surface at Day 90 as board defense failures.

The Five Board KPIs and the Named Workflows That Move Them

The five KPIs synthesize every named workflow from L1 through L5. The operator picks the five; the workflows are the means; the deltas are the commitment.

KPI 1 is platform-weighted booking percent. Stack: Avoca for missed-call recovery (100% answer rate, 39% cost-per-conversion reduction in the HL Bowman case), or Jobber AI Receptionist, or Housecall Pro AI Agents, or ServiceTitan Voice. Workflow: 4 p.m. CSR floor pull of AI-booked calls + daily missed-and-recovered report + weekly CSR audit from call recordings. Baseline 65%; 90-day target 78-82%. The single highest-ROI workflow in the program.

KPI 2 is platform-weighted Membership Penetration Rate. Stack: Rilla for ride-along coaching (18% close-rate lift documented, 30-40 virtual ride-alongs per manager per day), plus membership upgrade path coached at every touch, plus the tech's 30-second AI-generated membership pitch. Workflow: daily Rilla transcript review + weekly per-tech MPR scorecard + monthly platform MPR trend. Baseline 22%; 90-day target 32-40% trending to 35-50% by Day 180.

KPI 3 is platform-weighted financing close percent on $5K+ jobs. Stack: Wisetack / GreenSky / Synchrony tier matrix (Wisetack for fast service approvals, GreenSky for high-ticket replacement with deeper credit, Synchrony for branded revolving), plus skip-the-quote pre-approval (60-second soft-pull at the door), plus the AI-built homeowner credit stack (IRS 25C / 25D + state rebate + financing payment math), plus AI objection rebuttals at the kitchen table. Workflow: financing tier decision at dispatch + soft-pull at the door + AI-assembled credit stack + close cycle compression from 90 to 50 minutes. Baseline 14%; 90-day target 22-28% trending to 28-40%.

KPI 4 is platform-weighted recall percent. Stack: AI-powered RC&D triage with SLA tiers (recall 24-hour, callback 48-hour, warranty 5-business-day), root-cause clustering by tech / part / symptom / install date, plus customer recovery playbook. Workflow: Friday recall huddle with root-cause heatmap + manufacturer denial escalation + owner sign-off on disputed warranty cases. Baseline 5-7%; 90-day target under 4% trending to under 2%.

KPI 5 is platform-weighted GLSA ROAS (or RPL by source). Stack: CallRail Conversation Intelligence for source attribution, Ryze AI for closed-revenue-tied GLSA bid optimization, NiceJob / Podium AI Employee / Birdeye AI Employee for review automation, Hatch for stale-lead reactivation, Yelp AI for Yelp replies. Workflow: Monday marketing brief + Friday owner recap + weekly negative-lead disputes against the Ryze feedback loop + monthly Hatch cohort analysis. Baseline 3-4x ROAS; 90-day target 4-5x trending to 5-6x.

The five KPIs are the external promise; the named workflows are the internal means. The cascade from KPI to workflow to named owner to daily artifact converts 115 lessons into measurable EBITDA contribution.

The Capstone Synthesis — 115 Lessons in One Page

The plan is the synthesis of every prior lesson. L1's responsible AI discipline becomes Day 0 governance docs. L2's role-specific workflows (CSR rebuttals, dispatcher override discipline, tech job-notes, advisor financing pivots, marketing ROAS) become the named workflows behind each KPI. L3's end-to-end workflows (missed-call recovery, dispatch optimization, ride-along coaching, RC&D triage, marketing attribution, pricebook auto-update, fleet 179, trade-tax planning) become the platform deployment commitments. L4's 12-month roadmap and readiness audit become the multi-quarter context. L5's transformation framework, governance, org design, M&A workflow, synergy synthesis, hyperscale bid-prep, and AI-native dispatch become the platform-scale architecture.

The plan is the program's external accountability artifact. The operator walks into the Nexstar peer call, the franchise QBR, the PE operating partner monthly call, the lender covenant test, the cyber insurance renewal, the M&A diligence call — all with the same artifact: five KPIs, wave 1 site map, governance attestation, org-design execution, next 90-day commitment. Universal artifact because universal discipline.

The plan also positions the platform for the 2027-2030 hyperscale wave. The 300,000-electrician AI-data-center gap (CSIS 2026), HVAC engineers up 67%, robotics techs up 107% — 2026 reality, not future signal. The operator either positions for the wave or commits to an alternative-path EBITDA thesis. Both are defensible. Silence is not.

The CSR taught to rebut "I'm just calling for a price" in L2 becomes the booking percent KPI. The dispatcher taught to override Dispatch Pro with logged reason becomes dispatch yield. The advisor taught to assemble the IRS 25C credit stack becomes financing close. The service manager running the Friday recall huddle becomes recall percent. The marketing manager running the Monday brief becomes GLSA ROAS. The owner reading the 12-metric dashboard in L4 becomes the operator defending the five-metric board dashboard in L5. The plan is the diploma. Day 90 movement is the score. The next 90-day plan is the next semester. Compounding discipline across four quarters distinguishes the trades operators of 2027-2030 from those who bought AI tools without a plan.

Key Takeaways

• The L5 capstone is the 90-day plan, not a memo. Dated, named, metric-tied. Six Day 0 deliverables and three stage-gate checkpoints (Day 30 deployment, Day 60 lift, Day 90 board defense).
• Why 90 days, not 12 months: the L4 roadmap is single-shop annual; the L5 plan is platform-quarterly. 90 days aligns to the four-stage rollout framework, PE board cadence, franchise HQ QBR cadence, and lender covenant testing.
• Day 0 deliverables are six: executive summary; location-by-location pilot map with readiness tiers A/B/C; governance and policy docs (platform AI policy, vendor matrix, compliance attestation, breach response playbook); org-design (Director of AI Operations, Conversation QA Lead, Prompt Librarian, Data Center Account Director); data center positioning thesis (five questions); five-metric board dashboard.
• Day 30 is deployment discipline, not lift. Wave 1 sites in training / integration / first-baseline state. Exception log surfaces vendor engagement, owner stretch, operating-bench absorption. Board update: one slide.
• Day 60 is mid-cycle lift measurement. Avoca sites at 8-12% missed-call; Rilla teams at 4-8 close-rate points; Dispatch Pro at 6-10% RPT. Spread analysis is the first look at top vs. bottom quartile and informs the Day 90 wave 2 segmentation hypothesis.
• Day 90 is the board defense: five KPI movement against commitment; wave 1 site-level lift and spread analysis; governance attestation update; org-design execution; next 90-day plan. Operators who defend with spread analysis and documented dependencies compound credibility.
• The seven-question gate: (1) does it move a board KPI? (2) named owner + metric target? (3) dependency chain verified? (4) rollback trigger + customer-facing language? (5) vendor success manager committed in writing? (6) absorption load tolerable? (7) dashboard update for next board review?
• The five board KPIs map to named workflow stacks: booking % → Avoca / Jobber AI Receptionist / HCP AI Agents / ServiceTitan Voice; MPR → Rilla + membership upgrade path; financing close → Wisetack / GreenSky / Synchrony + IRS 25C/25D credit stack + skip-the-quote pre-approval; recall % → AI RC&D triage + root-cause clustering; GLSA ROAS → CallRail + Ryze AI + Hatch + NiceJob + Podium + Birdeye + Yelp AI.
• The plan synthesizes 115 lessons. L1 governance → Day 0 docs. L2 role workflows → named KPI workflows. L3 end-to-end workflows → platform deployment. L4 roadmap → multi-quarter context. L5 framework, governance, org design, M&A, synergy, hyperscale → platform architecture.
• The data center positioning thesis is a Day 0 deliverable because the 300K-electrician AI-data-center gap, 67% HVAC engineer growth, and 107% robotics-tech growth are 2026 reality. The operator either commits to the wave or commits to a defensible alternative-path EBITDA thesis. Silence is not defensible.
• The plan is the program's diploma. Day 90 movement is the score. The next 90-day plan is the next semester. Compounding discipline across four quarterly defenses distinguishes the operators of 2027-2030 from those who bought AI tools without a plan.