The audit produced the score. The roadmap produced the dated plan. The phase gates named the dependencies. Now the owner has the question every banker, peer-group call, and PE partner asks third: what does this actually cost, what is the year-one and year-two budget, and where is the payback math? This is the lesson that translates the 12-month roadmap into a defensible budget with line items, internal-time loading, contingency, and quarterly cash-out timing. Real 2026 vendor pricing — Avoca $1K-$3K/month, Rilla $200-$400/seat/month, Dispatch Pro add-on $200-$400/month, NiceJob / Podium AI Employee / Birdeye AI Employee $300-$600/month, Hatch $300-$600/month, CallRail Conversation Intelligence $145-$995/month depending on volume tier, Ryze AI or comparable GLSA bid platforms $500-$1,500/month, plus the in-platform AI add-ons (ServiceTitan Voice $300-$600/month, Titan Intelligence bundles, Sera AI scheduling fees). Plus internal time — the line item most owners forget — at $25-$45/hour blended cost across CSR, dispatcher, service-manager, advisor, and ops-manager attention. Total annual external spend for a 6-truck residential HVAC shop on the full Phase 1-4 stack: $25K-$60K depending on volume and tier choice. Total internal time: $15K-$35K. All-in: $40K-$95K year one, $35K-$80K year two as the team's attention drops from setup to operate. Payback target: 6-9 months on the full stack, with Phase 1 paying back at 60-90 days, Phase 2 at 90-150 days, Phase 3 at 120-180 days, Phase 4 at 270-360 days as the long-tail compounding kicks in. This lesson is the budget the owner walks into the banker conversation with, the line items the CFO needs for the COA mapping, and the contingency the peer group expects to see.

Why the Budget Is Built After the Audit and the Roadmap

Most owners try to build the budget first. They open a vendor pitch deck, see Avoca at $2K/month, add it to a spreadsheet, copy the line for Rilla and Hatch, total it, and call it a budget. That number is wrong in three directions and the owner does not know which way. It is wrong on tool selection (the audit may say Phase 1 launches month 4, not week 1, which shifts $24K of year-one Avoca cost off the year-one budget). It is wrong on internal time (the spreadsheet has no line for the CSR floor's 6 hours per week of pilot ramp time at week 1-3). And it is wrong on phasing (the spend looks like a flat monthly burn when it is actually backloaded — Phase 4 spend is double Phase 1 spend, and the cash-out timing matters at a shop running on a $50K-$200K working-capital line).

The audit-then-roadmap-then-budget sequence solves all three. The audit names which Phase 1 tool the shop actually qualifies for at week 1 vs. which needs 30-90 days of remediation. The roadmap names the month each tool goes live and the metric it targets. The budget then assembles the line items at the correct monthly cadence, layers in the internal-time loading per phase, adds the contingency band, and produces a quarterly cash-out schedule that the CFO or bookkeeper can map to the cash-flow forecast. The result is a budget that defends — the banker accepts it as a working-capital ask, the peer group accepts it as a benchmark, the PE partner accepts it as a synergy line, and the owner sleeps because the numbers actually tie out.

The budget also surfaces the payback math the audit and roadmap imply but do not compute. Phase 1's $35K of year-one Avoca-plus-ServiceTitan-Voice-plus-internal-time produces $280K of attributable booking-lift revenue at typical 38% gross margin — $106K of margin against $35K of spend, a 3x cash-on-cash year one and an 8x revenue payback. Each Phase has the same math the budget makes visible: external spend, internal time, attributable revenue lift, attributable margin, payback months. Without the budget, the owner argues payback verbally; with it, the owner shows the spreadsheet.

The Vendor Line Items — Real 2026 Pricing

The vendor stack for a 6-truck residential HVAC shop running ServiceTitan as the FSM platform, executing the full Phase 1-4 roadmap, lists nine to twelve external line items. The pricing bands below reflect 2026 published or negotiated tier rates as of mid-2026.

Phase 1 vendors (months 1-3). Avoca voice agent at $1K-$3K/month depending on call volume tier (a 6-truck shop running 1,800-2,400 monthly inbound calls lands at $1,800-$2,400/month). ServiceTitan Voice or Titan Intelligence call-summary AI add-on at $300-$600/month (shops that pick the in-platform bundle instead of Avoca; some shops run both, treating Avoca as the bolt-on for after-hours overflow). Phone-system upgrade or Avoca-compatible SIP carrier line at $100-$300/month incremental. Phase 1 monthly external: $2,200-$3,300; Phase 1 quarterly: $6,600-$9,900.

Phase 2 vendors (months 4-6). ServiceTitan Dispatch Pro at $200-$400/month add-on (or Sera profit-aware scheduling at comparable pricing for Sera shops, or FieldEdge AI scheduling for FieldEdge shops). Service-manager scorecard tool or dashboard build-out at $0-$500/month (most shops use ServiceTitan native reporting; some buy a layered BI tool like Servicetrades, Bridge, or Profit Rhino for $200-$500/month). Phase 2 monthly external: $200-$900 incremental on top of Phase 1; Phase 2 quarterly incremental: $600-$2,700.

Phase 3 vendors (months 7-9). Rilla virtual ride-alongs at $200-$400/seat/month — for a 6-truck shop with 2 Comfort Advisors and 1 service manager seat, that is 3 seats at $600-$1,200/month total. ResponsiBid AI quoting and proposal generation at $150-$400/month per shop. AI-powered RC&D triage tools or ServiceTitan native classifier setup time (one-time $1K-$3K setup, ongoing $0-$200/month). Phase 3 monthly external: $750-$1,800 incremental on top of Phases 1 and 2; Phase 3 quarterly incremental: $2,250-$5,400.

Phase 4 vendors (months 10-12). CallRail Conversation Intelligence at $145-$995/month depending on volume tier (a 6-truck shop with 1,800-2,400 monthly inbound lands at the $290-$495/month tier). Ryze AI or comparable GLSA bid management platform at $500-$1,500/month depending on monthly ad spend (a $4,400/week GLSA spend = $19K/month places the shop at $750-$1,200/month). NiceJob review acquisition at $99-$299/month. Podium AI Employee 2026 review-response and lead-capture chat at $300-$600/month. Birdeye AI Employee 2026 review-response and sentiment monitoring at $300-$600/month (most shops pick one of NiceJob, Podium, or Birdeye, not all three; the line item represents the chosen vendor). Hatch stale-lead reactivation at $300-$600/month. Phase 4 monthly external: $1,800-$3,400 incremental on top of Phases 1-3; Phase 4 quarterly incremental: $5,400-$10,200.

Year-one external vendor total at the low end: $25K-$30K (lighter Phase 4 selections, lower-tier Phase 1). Year-one at the mid: $40K-$50K (full Phase 4 stack with mid-tier Phase 1). Year-one at the high end: $55K-$75K (full stack with top tiers and added bolt-ons). A 6-truck residential HVAC shop typically lands at the $40K-$50K mid-band on year one. Year two drops 10-15% as setup fees fall off and volume-tier renegotiation lands annual discounts at the 12-month mark.

The Internal Time Line Item Most Owners Miss

The vendor line items are visible. The internal time is invisible until it is too late. Every AI tool requires team attention during setup, ramp, and ongoing operation; the cost is real even when it does not produce an invoice. The budget must include it explicitly or the year-one numbers miss the actual investment by 30-50%.

Phase 1 internal time. CSR floor pilot ramp at 6 hours per CSR per week for weeks 1-3, then 1-2 hours per week ongoing — 3 CSRs × 6 hours × 3 weeks plus 3 CSRs × 1.5 hours × 49 weeks = 275 hours year one at $25/hour blended cost = $6,900. Office manager or operations manager daily Avoca review at 30 minutes per day × 5 days × 50 weeks = 125 hours at $35/hour = $4,400. Owner weekly Phase 1 review at 30 minutes per week × 50 weeks = 25 hours at $75/hour blended = $1,875. Phase 1 internal time total year one: $13,000-$15,000.

Phase 2 internal time. Dispatcher daily override discipline at 15 minutes per day × 5 days × 40 weeks (starting month 4) = 50 hours at $30/hour = $1,500. Service manager weekly RPT-by-tech review and dispatch-yield analysis at 1 hour per week × 40 weeks = 40 hours at $40/hour = $1,600. Owner Friday dispatch-yield trend review at 15 minutes per week × 40 weeks = 10 hours at $75/hour = $750. Phase 2 internal time year one: $3,500-$4,500.

Phase 3 internal time. Service manager daily Rilla transcript review at 30 minutes per day × 5 days × 30 weeks (starting month 7) = 75 hours at $40/hour = $3,000. Service manager weekly coaching with each advisor — 2 advisors × 30 minutes × 30 weeks = 30 hours at $40/hour = $1,200. Comfort Advisor lead close-rate scorecard review at 30 minutes per week × 30 weeks = 15 hours at $50/hour = $750. Office manager AI RC&D classifier audit at 30 minutes per week × 30 weeks = 15 hours at $35/hour = $525. Phase 3 internal time year one: $4,500-$6,000.

Phase 4 internal time. Marketing manager weekly CallRail review at 1 hour × 12 weeks = 12 hours at $45/hour = $540. Marketing manager monthly Hatch segment design and review at 2 hours × 3 months = 6 hours at $45/hour = $270. Office manager NiceJob / Podium / Birdeye daily 60-second skim — 5 minutes per day × 5 days × 12 weeks = 5 hours at $35/hour = $175. Owner daily 12-metric dashboard read at 8 minutes per day × 5 days × 12 weeks = 8 hours at $75/hour = $600. Phase 4 internal time year one: $1,500-$2,500.

All-Phases internal time year one: $22,500-$28,000. Add 15-20% contingency for higher-than-projected pilot ramp time on at least one Phase: $25,000-$33,000. Year two internal time drops to $15,000-$20,000 as the routines become habit and the setup overhead falls away.

Contingency and Quarterly Cash-Out Timing

A budget without contingency is aspirational. A budget with contingency defends. The 2026 trades-AI budget contingency band is 15-25% of the external vendor spend plus 10-15% of the internal time loading — based on documented overruns across shops that have published their year-one execution.

The three largest contingency drivers. One: integration or data hygiene work surfaces during Phase 1 (typical $2K-$8K of CSR or office-manager time to clean equipment records, address fields, or membership tagging that the audit identified as weak). Two: vendor pilot extension or setup fees beyond the initial quote (typical $1K-$5K when a tool requires custom integration or a SIP carrier upgrade). Three: training overrun on Phase 2 or Phase 3 when a service manager, advisor, or dispatcher needs more ramp time than projected (typical $2K-$10K of additional internal hours).

The quarterly cash-out timing matters more than the annual total for shops running on a working-capital line. Phase 1 cash-out concentrates in Q1: $6,600-$9,900 of external vendor spend plus $4,000-$5,000 of internal time plus $1K-$3K of setup fees = $11K-$18K cash-out in Q1. Phase 2 adds incremental in Q2: $600-$2,700 external plus internal time = $2K-$5K incremental in Q2. Phase 3 adds incremental in Q3: $2,250-$5,400 external plus $4,500-$6,000 internal = $7K-$11K incremental in Q3. Phase 4 adds the largest incremental in Q4: $5,400-$10,200 external plus internal time = $7K-$12K incremental in Q4. Year-one total cash-out: $27K-$46K across the four quarters with a clear backloading toward Q3-Q4 as the full stack lights up.

The cash-out timing creates an important budgeting nuance. A shop running a $50K working-capital line can absorb the year-one AI investment without a separate financing event. A shop running a $20K line needs to either stage Phase 3 and Phase 4 over two fiscal years or arrange supplemental working-capital before Q3. The audit-roadmap-budget chain makes the financing conversation deliberate; the owner who plans the cash-out timing avoids the surprise mid-year cash crunch that derails year-one execution at 30-40% of shops attempting full Phase 1-4 in 12 months.

The Payback Math and the Defense of the Budget

The budget defends in front of a banker, peer group, or PE partner only if it ties to attributable margin. The payback math is the second page of the budget deck and ties every spend line item to a revenue or margin line item.

Phase 1 payback. External spend: $25K-$30K year one. Internal time: $13K-$15K. Total Phase 1: $38K-$45K. Attributable revenue: $200K-$400K of booking-lift recovered revenue at the 65%-to-80% booking lift on 1,800-2,400 monthly inbound calls × 12 months. Take the midpoint: $300K revenue at 38% gross margin = $114K gross margin. Phase 1 payback: $38K-$45K cost against $114K gross margin = 2.5-3x cash-on-cash year one, payback at 4-5 months of operation.

Phase 2 payback. External spend incremental: $2.4K-$10.8K year one (4 months of operation × $600-$2,700 monthly). Internal time: $3.5K-$4.5K. Total Phase 2 incremental: $6K-$15K. Attributable revenue: 12-18% RPT lift on $1,900/day baseline × 6 trucks × 250 working days × 60% of year-one = $164K-$246K. Take midpoint: $205K revenue at 60-70% gross margin on incremental marginal-truck-day revenue = $130K gross margin. Phase 2 payback: $6K-$15K against $130K = 9-22x cash-on-cash on incremental year-one Phase 2 spend.

Phase 3 payback. External spend incremental: $6.7K-$16.2K year one (3 months of operation × $2,250-$5,400 monthly). Internal time: $4.5K-$6K. Total Phase 3 incremental: $11K-$22K. Attributable revenue: close-rate lift of 8-14 points on $5K+ proposals plus MPR lift plus recall reduction = $250K-$400K. Take midpoint: $325K revenue at 40-45% gross margin on the close-rate-driven mix = $135K gross margin. Phase 3 payback: $11K-$22K against $135K = 6-12x cash-on-cash on Phase 3 incremental.

Phase 4 payback. External spend incremental: $5.4K-$10.2K year one (3 months of operation × $1,800-$3,400 monthly). Internal time: $1.5K-$2.5K. Total Phase 4 incremental: $7K-$13K. Attributable revenue: GLSA ROAS lift 30-50% on $4,400/week spend × 3 months = $40K-$60K of recovered GLSA revenue plus Hatch reactivation revenue $30K-$60K plus owner-dashboard-driven optimization gains. Take midpoint: $90K revenue at 38% gross margin = $34K gross margin. Phase 4 payback: $7K-$13K against $34K = 3-5x cash-on-cash on Phase 4 incremental; long-tail compounding lifts Phase 4 payback to 5-8x by month 18.

Year-one all-in. External spend: $40K-$50K. Internal time: $25K-$33K. Total year-one cost: $65K-$85K. Attributable revenue: $900K-$1.3M. Attributable gross margin: $370K-$520K. Year-one cash-on-cash: 4-7x; year-one revenue payback: 11-20x. Payback timing: 6-9 months on the full stack with Phase 1 paying back at month 4-5, Phase 2 by month 8, Phase 3 by month 10, Phase 4 by month 14-18.

How the Budget Changes for Different Shop Profiles

The 6-truck residential HVAC shop budget anchors the lesson. The numbers shift predictably for different shop profiles based on truck count, FSM platform, trade mix, and Phase scope ambition.

1-3 trucks on Jobber or Housecall Pro. Phase 1: bundled in-platform AI (Jobber AI Receptionist at $50-$200/month incremental or HCP AI Agents at $50-$300/month) instead of Avoca. Phase 2-3 deferred to year 2 (single-truck shop has no dispatch AI ROI; 2-3 trucks barely justify the discipline). Phase 4: lighter — CallRail at $145/month, NiceJob at $99-$149/month, Hatch deferred. Year-one external: $4K-$12K. Internal time: $5K-$10K. Total year-one: $9K-$22K. Payback: 3-6 months on a $50K-$150K attributable revenue lift.

4-15 trucks on ServiceTitan or Sera. Phase 1-4 on schedule. Avoca primary plus ServiceTitan Voice as overflow secondary. Full Phase 4 stack including Ryze AI bid management. Year-one external: $50K-$80K. Internal time: $25K-$35K. Total year-one: $75K-$115K. Payback: 6-9 months on $900K-$1.5M attributable revenue lift.

16-40 trucks. Phase 1-4 accelerated to 9-month execution. Multiple Comfort Advisor seats on Rilla (4-8 seats × $300/month = $1,200-$2,400/month vs. the 3-seat 6-truck case). Dedicated ops or AI lead role at $40K-$80K/year — the internal time loading shifts from incumbent staff time to a named hire. Year-one external: $80K-$130K. Internal time including the AI lead: $80K-$140K. Total year-one: $160K-$270K. Payback: 5-8 months on $2.5M-$5M attributable revenue lift.

40+ trucks or multi-shop. Platform-mandated vendor stack (most PE platforms — Wrench Group, Authority Brands, Apex Service Partners, Sila Services, Path Light Pro, Redwood Services — standardized on Avoca cross-portfolio, Rilla cross-portfolio, ServiceTitan or BuildOps cross-portfolio in 2025-2026). Per-location budget allocations from HQ; per-location internal time loading at $50K-$120K/year. Per-location year-one all-in: $150K-$300K. Per-location payback: 6-9 months on $3M-$8M location revenue lift.

The budget structure stays consistent across profiles — vendor line items, internal time loading, contingency, quarterly cash-out, payback math. The numbers scale linearly with truck count; the bands scale by FSM platform and Phase ambition.

The Budget as the Banker and PE Conversation Artifact

The budget is not an internal document. It is the artifact the owner presents to the working-capital banker, the peer-group benchmark call, the franchise HQ technology assessment, or the PE quarterly business review. The artifact has three pages.

Page 1: the budget summary. One table. Twelve rows (one per month). Five columns: month, phase, external vendor spend, internal time loading, total monthly cash-out. Total row at the bottom: year-one total. The page reads in 90 seconds and answers "what does year one cost."

Page 2: the payback math. One table. Four rows (one per phase). Six columns: phase, external spend incremental, internal time, total cost, attributable revenue, attributable margin, payback months. Total row: year-one all-in. The page reads in 2 minutes and answers "what does year one return."

Page 3: the contingency and risk register. Three sections. Vendor risk (volume-tier escalations, integration setup overrun, vendor churn during pilot). Internal-time risk (CSR floor pushback extending ramp, service manager turnover during Phase 2-3, owner attention diverted by a service crisis). Cash-flow risk (working-capital line constraints in Q3-Q4 backloading). Each risk has a probability band and a mitigation. The page reads in 4 minutes and answers "what could go wrong and how do you handle it."

The three-page budget deck is the L4 capstone artifact and the peer-group benchmark artifact. The owner who walks in with the deck — page 1 ties to the roadmap, page 2 ties to the audit's projected metric movement, page 3 names the risks — has the conversation that ends with "what do you need from us to support it." The owner who walks in with "we are budgeting roughly $50K for AI this year" has the conversation that ends with "send me a proper budget when you have one." Same shop. Same intent. Different outcomes. The budget is the artifact.

Key Takeaways

• The audit produced the score; the roadmap produced the plan; the budget produces the defensible number. Sequence matters. A budget built before the audit and roadmap is wrong on tool selection, internal time, and phasing.
• Year-one external vendor spend for a 6-truck residential HVAC shop: $25K-$60K. Phase 1 (months 1-3): $2,200-$3,300/month — Avoca, ServiceTitan Voice, phone-system upgrade. Phase 2 (months 4-6): $200-$900/month incremental — Dispatch Pro / Sera / FieldEdge AI scheduling. Phase 3 (months 7-9): $750-$1,800/month incremental — Rilla seats, ResponsiBid, RC&D triage tools. Phase 4 (months 10-12): $1,800-$3,400/month incremental — CallRail, Ryze AI, NiceJob / Podium / Birdeye, Hatch.
• Internal time loading is the line item most owners miss. $22K-$28K year one across CSR floor pilot ramp, ops-manager daily Avoca review, dispatcher override discipline, service-manager Rilla transcript review, marketing-manager CallRail review, owner dashboard time. Drops to $15K-$20K year two as routines become habit.
• Contingency band: 15-25% of external spend plus 10-15% of internal time. Three largest drivers: data-hygiene work surfaces during Phase 1, vendor pilot extension or setup fees, training overrun on Phase 2 or Phase 3.
• Quarterly cash-out backloads to Q3-Q4. Q1: $11K-$18K. Q2: $2K-$5K incremental. Q3: $7K-$11K incremental. Q4: $7K-$12K incremental. Year-one total cash-out: $27K-$46K. A $50K working-capital line absorbs the investment; a $20K line requires staging across two fiscal years.
• Payback math by phase. Phase 1: 2.5-3x cash-on-cash year one, 4-5 months. Phase 2: 9-22x on incremental. Phase 3: 6-12x on incremental. Phase 4: 3-5x year one, 5-8x by month 18 on long-tail compounding.
• Year-one all-in for a 6-truck shop: $65K-$85K cost against $900K-$1.3M attributable revenue and $370K-$520K attributable gross margin. 4-7x year-one cash-on-cash; 11-20x revenue payback; full-stack payback at 6-9 months.
• Budget scales linearly with truck count. 1-3 trucks: $9K-$22K year one. 4-15 trucks: $75K-$115K. 16-40 trucks: $160K-$270K with dedicated AI lead role. 40+ trucks: $150K-$300K per location with PE-platform-mandated vendor stack.
• The budget is a three-page artifact. Page 1: monthly summary (12 rows, 5 columns). Page 2: payback math (4 rows, 6 columns). Page 3: contingency and risk register. The deck reads in 7 minutes and defends in front of a banker, peer group, franchise HQ, or PE partner.
• The budget is the L4 capstone artifact. A budget tied to the audit and the roadmap converts AI strategy from narrative to evidence. Owners who walk into a peer-group or PE call with the deck get "what do you need to support it"; owners without get "send me a proper budget when you have one."