Every residential service tech has the same 90-second decision in front of a 19-year-old condenser on a Tuesday afternoon: tell the homeowner to repair, tell them to replace, or hand off to a Comfort Advisor for tomorrow's three-option sit-down. The tech who freelances the conversation gets one of three things wrong every time — over-recommends repair on a unit clearly past its economic life, under-recommends replacement because they want the smaller-ticket-and-go, or pushes replacement so hard the homeowner shuts down and calls a competitor. None of these are bad techs. They are techs without a structured talk-track. AI fixes the freelance. The tech opens a prompt template, dictates five inputs — equipment age, repair cost, customer goals, utility rebate, financing payment — and the AI returns a three-option (repair / partial / replace) talk-track the tech reads off the tablet in 90 seconds. Mid-tier option closes 52-58% of the time in shops running this discipline. Average ticket on the conversation lifts $320-$680 because the partial option (the unit-only or the system-with-deferred-add-ons) captures the homeowner who was going to refuse the full replacement but would commit to the middle path. This lesson is the five-input prompt build, the three-option output template, the good/better/best presentation discipline behind it, the named pricing-model AI tools the shop runs, and the 30-day deployment that lifts close rate 6-12 points on the diagnostic-to-replacement pivot.

Why the Repair vs. Replace Conversation Is the Single Highest-Stakes Tech Moment

The repair vs. replace pivot is the single highest-revenue moment of the tech's day. A service ticket repairs at $400-$900 average. A replacement closes at $11K-$28K residential, with commercial drifting toward $50K-$250K. The pivot from one to the other is where the dollar delta lives, and the tech is the only role in the shop standing next to the equipment when it happens. The Comfort Advisor closes the replacement, but the tech opens the door — the tech's recommendation in the first 6 minutes of the conversation determines whether the Advisor walks into the kitchen tomorrow with a warm homeowner who has been told the unit is past life, or a cold homeowner who has been told "let's see if we can fix it for now."

Service MVP / Joe Crisara built a 30-year consulting business on this single pivot. The Power Selling Pros library, ResponsiBid's good/better/best framework, ServiceTitan's pricebook-driven three-option presentation, and the Comfort Advisor scorecards all converge on the same shape: present three options, let the homeowner choose, do not freelance. What AI changes in 2026 is the speed at which the tech can produce a customer-specific three-option talk-track on the tablet at the curb, without dragging the conversation into the 25-minute pricebook-flip that the homeowner experiences as a sales pitch. The prompt is built once, the talk-track is generated in 14 seconds, and the tech reads it off the tablet like a script that sounds like the tech because the prompt is calibrated to shop voice.

The published industry data anchors the lift. Built on Tenth's 2026 close-rate study found that techs running a structured three-option pivot lifted diagnostic-to-replacement-quote conversion 18-24 points over freelance-pivot techs, and the mid-tier close rate on those structured presentations landed at 52-58% — meaning the partial option (the unit-only path, or the system-with-deferred-add-ons path) captured roughly half the homeowners who would have refused the full replacement but were ready to commit to the middle path. The good/better/best framework also lifts the better-tier-vs-best-tier upgrade close by 8-14 points because the AI computes the per-month payment delta and the rebate-stack delta in real time at the curb, removing the friction of "let me check on that and call you back."

The Five-Input Prompt the Tech Dictates Curbside

The talk-track lives downstream of five specific inputs the tech dictates into the prompt template after the diagnostic is complete. The inputs are short — one phrase each — and the tech speaks them in 30-45 seconds total. The AI's output quality scales with the completeness of these five inputs; skipping one produces a talk-track that drifts toward generic and the tech ends up freelancing.

Input One: Equipment Age

Make, model, serial-derived age, and the practical economic-life context. "Trane XR15 outdoor unit, manufactured June 2008, paired with a 2008 American Standard 80% AFUE upflow furnace — system is 18 years old, the average service life for this product line is 12-15 years, this unit is past it." The age input does two things: it anchors the repair-vs-replace recommendation to actual equipment life expectancy, and it gives the AI the structural data to compute the "remaining life cost-per-year" math the homeowner will hear in the talk-track. Without the age input, the AI defaults to generic "your system is getting older" language the homeowner pattern-matches as upsell. With it, the AI produces "the average service life on this make is 12-15 years; yours is 18, which means every repair from here is rented time — and the math is roughly $90-$140 per remaining month of life."

Input Two: Repair Cost

The specific repair dollar amount, with the parts breakdown the homeowner will see on the invoice. "Repair is $1,247 — that's a new compressor contactor at $187, low-voltage transformer at $94, capacitor at $145, the labor at $621 because the contactor is mounted under the coil and requires recovering the refrigerant, and the system charge at $200." Specific dollar amount and the labor-driver explanation. The AI uses the repair cost to compute the 50% rule (if repair is >50% of replacement cost, replace) and the cost-per-remaining-month-of-life math. Without the specific repair cost, the talk-track produces hand-wavy "repair will cost a few hundred dollars" language the homeowner does not commit to.

Input Three: Customer Goals

What the homeowner said about what they want — and what they did not say. "Husband said 'we want this to last us through retirement, we're 7 years out.' Wife said 'efficiency matters because the bills last winter were brutal.' Both mentioned they have grandkids visiting in July and 'we cannot go without cool again.' Husband is on Medicare, wife is on a small annuity — they are not on a fixed shoestring but they pay attention to monthly costs." Customer goals input drives whether the talk-track leads with cost-per-month, with total, with rebate stack, or with comfort delta. The Advisor reading the same field tomorrow tunes the kitchen-table presentation off it, but the tech's tablet talk-track today opens the conversation calibrated to the homeowner's actual words.

Input Four: Utility Rebate

The current rebate landscape for this homeowner's ZIP code, equipment type, and efficiency tier. "TVA EnergyRight rebate is $750 on a heat-pump replacement at 16+ SEER2. Local gas utility offers $400 on a 96% AFUE furnace tied to the same install. State of Tennessee weatherization-assistance program is open and the homeowner qualifies based on county income table — that's another $1,200 toward the panel-upgrade portion." Specific rebate amounts and source utilities, computed against the homeowner's ZIP. The AI does not invent rebates; the rebate amounts get loaded into the prompt template each quarter from the shop's rebate-tracking sheet (or from a per-territory rebate-lookup AI like the one Sera AI ships in their 2026 release, or the Section 25C / 25D credit-stack tool the Comfort Advisor uses in Ch 5). Without the rebate input, the talk-track leaves $1,500-$4,000 of homeowner-incentive on the floor, and the close rate on the mid-tier and best-tier options drops 8-14 points.

Input Five: Financing Payment

The estimated monthly payment for each option, based on the shop's primary financing tier matrix. "Wisetack 0% for 24 months on the repair-only option lands at $52/month. GreenSky 84-month at 8.99% on the partial-replacement (unit only) lands at $187/month. GreenSky 120-month at 7.49% on the full replacement with panel upgrade and surge-protect lands at $238/month." The payment numbers come from the shop's actual financing portal — the tech enters the total estimated cost into the Wisetack or GreenSky soft-pull at the door (which was triggered when the dispatch went out, per Ch 5 Lesson 3), and the approved tier comes back with the monthly payment math. The AI does not invent APR or payment. The constraint in the prompt is explicit: "Do not invent APR. Use only the payment figures I dictate from the financing portal." Reg Z exposure applies — the owner has personal liability for any payment language the AI produces in the homeowner's record.

The Three-Option Output Template

The AI's output is structured into three options — repair, partial, replace — that the tech reads off the tablet in 90 seconds. The structure is non-negotiable and matches the good/better/best framework the Comfort Advisor uses at tomorrow's kitchen-table sit-down, so the conversation continuity from tech-tablet to Advisor-presentation is preserved.

Option One: Repair

The keep-the-current-unit-running option. The AI talk-track names the repair cost, the parts being replaced, the warranty window on the repair (typically 90 days on the part, 1 year on the labor), and the explicit framing of remaining life. Example: "Option one is the $1,247 repair. We replace the contactor, the transformer, and the capacitor, and you get 90 days on the part and 1 year on the labor. The honest truth is your system is 18 years old, and the average life on this make is 12-15 years — so this repair gets you through the summer, probably through the winter, but the next failure is likely to be the compressor or the coil, and at that point the repair cost runs $2,800-$4,500 and we are back at this same conversation in 8-14 months. The math on this option is roughly $90-$140 per remaining month of life, depending on how long the unit holds." Honest framing, no soft-pedaling, anchored to the equipment-age input.

Option Two: Partial (the Mid-Tier)

The unit-only replacement or the system-with-deferred-add-ons option. This is the option that closes at 52-58% in disciplined shops, because it captures the homeowner who refuses the full system but commits to the middle path. Example: "Option two is the partial-replacement at $11,400. We replace the outdoor condenser with a 16-SEER2 unit, leave the existing furnace in place because it's still serviceable, and add the matching evaporator coil. You get the 10-year parts manufacturer warranty and a 2-year shop labor guarantee. With the $750 TVA EnergyRight rebate that brings net cost to $10,650. Financing at GreenSky 84 months at 8.99% is $187/month. This option gets you the cooling reliability you said you wanted before the grandkids visit, and you defer the furnace decision 4-7 years to when it actually fails." The partial option is also where the tech's job-notes from Lesson 1 (customer-mood and financing-posture fields) calibrate the framing — the AI reads those fields and tunes the partial option to the homeowner's actual stated priorities.

Option Three: Replace (the Full System)

The full system-and-add-ons option. The AI talk-track presents this with the rebate stack fully assembled (TVA EnergyRight + utility gas rebate + state weatherization + Section 25C credit), the efficiency delta vs. the partial option, and the comfort delta (two-stage vs. single-stage, variable-speed vs. PSC, communicating thermostat). Example: "Option three is the full system at $18,950 — that's the 18-SEER2 heat pump, the matching variable-speed air handler, the panel upgrade for the new amp draw, the smart thermostat, and the surge-protect. Rebate stack is $750 TVA + $400 gas + $1,200 state weatherization + $2,000 federal Section 25C credit = $4,350. Net cost $14,600. GreenSky 120 months at 7.49% is $238/month. You get a 25-30% utility-bill reduction over the current system and a 10-year complete-system warranty." The replace option is also where the Comfort Advisor's sit-down tomorrow lives — the tech is not closing the replacement at the curb; the tech is opening the door and handing off a warm homeowner to the Advisor who walks in tomorrow.

The 90-Second Read

The tech reads all three options to the homeowner in 90 seconds, off the tablet, sitting at the kitchen table or standing at the equipment, not from memory. The honest framing (this is what each option costs, this is what each option gets you) is what differentiates this from a sales pitch. The homeowner picks the option, or asks the Comfort Advisor question, or asks for time. The tech then either runs the repair (option one), books the Comfort Advisor (option two or three), or schedules the call-back (homeowner needs time). The 90-second read replaces the 25-minute freelance the average untrained tech runs, and the close rate at the diagnostic-to-quote pivot lifts 18-24 points.

The Good/Better/Best Presentation Discipline Behind the Talk-Track

The three-option template is not arbitrary — it matches the good/better/best framework that closes 52-58% of mid-tier choices in published trades sales research. The discipline behind why the framework works runs deeper than "give the customer choices."

The good/better/best framework leverages three well-documented psychological dynamics. First, decision anchoring: the homeowner's reference price gets set by the highest option (the $18,950 full replacement), making the partial ($11,400) feel reasonable by comparison. Second, choice architecture: three options activate the homeowner's evaluative reasoning rather than triggering the binary "yes-no" reflex that the single-option pitch produces. Third, autonomy preservation: the homeowner choosing the mid-tier feels they exercised judgment rather than being upsold, which dramatically lifts post-purchase satisfaction and downstream review rates.

ResponsiBid, ServiceTitan Pricebook, FieldEdge Proposal Builder, and the AI proposal tools in Ch 5 all converge on this framework. What AI in 2026 changes is the customization speed. Pre-AI, the tech presented either a single-option pitch or a generic three-option pricebook-flip. AI-generated talk-tracks customize the three options to the specific homeowner's equipment age, repair cost, goals, rebate landscape, and financing approval — all in 14 seconds — which lifts the mid-tier close from the pricebook baseline of 38-44% to the disciplined-AI baseline of 52-58%. The 14-point lift is the difference between a $12K and a $14K average ticket on the repair-vs-replace pivot conversation.

The discipline also requires that the tech does not lead with the mid-tier or argue toward the mid-tier. The AI presents three options honestly; the homeowner picks. Techs who slip into "if I were you, I'd go with the middle one" language reset the choice-architecture dynamic into a single-option pitch and the close rate collapses back to baseline. The Tuesday huddle reinforces the discipline: present three, let them pick, do not advocate.

The Tool Stack and the Shop Templates

The talk-track lives in a saved prompt template inside the shop's existing AI tooling. Five major platforms support this workflow in 2026.

ServiceTitan Titan Intelligence with Pricebook integration is the largest deployment surface. Titan Intelligence pulls the three-option pricing from the shop's calibrated Pricebook (the same source the Comfort Advisor uses), so the tech's tablet talk-track and the Advisor's sit-down are wired to identical numbers. Calibration: 6-8 weeks. ResponsiBid Pro is the specialty proposal builder for residential trades — the 2026 AI module integrates the five-input curbside dictation into both the talk-track and the formal kitchen-table proposal from one set of inputs. Its strength is the rebate-stack and financing-tier integration; ships with Section 25C / 25D credit logic prewired. Calibration: 2-4 weeks.

Housecall Pro AI with Estimate AI ties tech-side three-option generation to the customer-facing estimate document. Lightweight, fits 1-6-truck shops, calibrates in 2 weeks. Sera AI Sales Companion bundles the repair-vs-replace prompt with the dispatch-yield optimization layer — the talk-track output also flags the Advisor sit-down booking into the next-day's auto-route logic at the optimal slot. Lifts diagnostic-to-Advisor conversion 12-18% over standalone generation. Claude Projects or ChatGPT Team is the 48-hour bridge: workspace with template saved as system prompt and rebate/financing data updated quarterly. Total time per talk-track 3-4 minutes vs. 90 seconds with native.

The Verification, the Verify Checkpoints, and the Failure Modes

Every three-option talk-track gets the 30-second verify pass before the tech reads it to the homeowner. Job-notes from Lesson 1 cover the equipment-and-condition record; the repair-vs-replace talk-track adds three new verify checkpoints.

Rebate-stack verification. Every named utility rebate, state credit, and federal Section 25C / 25D figure must be either dictated by the tech from the shop's rebate-tracking sheet or pulled from a calibrated rebate-lookup tool. The AI must not invent rebate amounts. The consequence of fabrication: the homeowner commits to an option with a rebate that does not materialize and the shop eats the delta or loses the customer.

Financing-payment verification. Wisetack or GreenSky monthly payment must come from the shop's actual financing portal, not AI computation. APR fabrication is Reg Z exposure. 50% rule verification. If repair cost exceeds 50% of partial-replacement cost, the AI flags the repair option as economically unfavorable. The tech verifies the rule was applied — a $4,800 repair on a $9,000 partial should be flagged; a $1,247 repair on an $11,400 partial is well under threshold and presents honestly.

Three failure modes are specific to this lesson. First, the "lead with mid-tier" failure: the tech argues toward the partial rather than presenting all three honestly. Close rate collapses because the choice-architecture dynamic resets. Fix: Tuesday huddle reinforcement and random call-recording audits. Second, the "skip the customer-goals input" failure: AI talk-track defaults to generic framing; close rate drops 8-14 points. Fix: huddle practice and a 30-second customer-goals dictation built into the prompt UI. Third, the "stale rebate data" failure: rebate amounts in the prompt template are 6+ months out of date. Fix: quarterly rebate-sheet update tied to the first-Monday operations huddle, owned by the service manager or marketing manager.

The 30-Day Deployment and the Metrics That Prove the Lift

Week 0: service manager and lead tech draft the five-input prompt template and three-option output schema. Calibrate against last quarter's 10 highest-revenue replacement quotes and 10 highest-frequency repair tickets, ensuring the AI's output matches the shop's actual Pricebook and Advisor proposal language. Calibration: 2-3 hours. Week 1: pilot with 2-3 senior techs. Coaching points: dictation completeness, tablet-read discipline (read off the tablet, don't improvise), do-not-advocate discipline. Weeks 2-3: full-truck rollout. Tuesday huddle reinforces present-three-honestly. The 4 p.m. dispatcher review surfaces freelance cases. Week 4: integration with the Comfort Advisor desk — sit-down close rates lift 6-12 points where the tech's three-option preview matched the Advisor's presentation continuity.

Five numbers track the rollout. Diagnostic-to-quote pivot rate: baseline 32-44%, target 56-68%. Mid-tier close rate: baseline 38-44%, target 52-58%. Average ticket on the pivot: baseline $4,200-$6,800, target $4,520-$7,480 (the $320-$680 lift). Comfort Advisor sit-down close rate: baseline 28-36%, target 34-48% (the 6-12 point lift from continuity). Rebate-stack capture rate: percentage of talk-tracks that present a fully assembled rebate stack. Target 90%+. Stale-rebate-data drift surfaces in this metric first.

Key Takeaways

• The repair vs. replace pivot is the single highest-revenue moment of the tech's day. The dollar delta between a $400-$900 repair and an $11K-$28K replacement lives in the first 6 minutes of the conversation. AI eliminates the freelance and produces a 90-second three-option talk-track the tech reads off the tablet.
• Five inputs drive the talk-track: Equipment Age, Repair Cost, Customer Goals, Utility Rebate, Financing Payment. Skipping any one drops close rate 8-14 points. The tech dictates all five in 30-45 seconds; the AI generates the three-option output in 14 seconds.
• The three-option structure is non-negotiable: Repair, Partial, Replace. Same order every time. Matches the good/better/best framework the Comfort Advisor uses at the kitchen-table sit-down, preserving conversation continuity from curb to kitchen.
• Mid-tier (partial) closes at 52-58% in disciplined shops. The partial option captures the homeowner who refuses the full system but commits to the middle path — typically unit-only replacement with deferred furnace, or system-only with deferred panel upgrade.
• Good/better/best leverages three psychological dynamics: decision anchoring (highest option sets reference), choice architecture (three options activate evaluative reasoning), autonomy preservation (mid-tier choice feels like exercised judgment, not upsell).
• The tool stack: ServiceTitan Titan Intelligence + Pricebook, ResponsiBid Pro (2-4 week calibration), Housecall Pro AI + Estimate AI, Sera AI Sales Companion (deepest integration to dispatch-yield), or Claude Projects / ChatGPT Team as 48-hour bridge.
• Three new verify checkpoints specific to this conversation: rebate-stack (every utility, state, and federal credit), financing-payment (Wisetack or GreenSky from actual portal, never AI-computed), 50% rule (repair vs. partial cost ratio). All three are Reg Z and warranty exposures if fabricated.
• Three failure modes to defend against: lead-with-mid-tier (collapses choice architecture), skip-customer-goals (defaults to generic framing), stale-rebate-data (the talk-track promises rebates that no longer exist). Quarterly rebate-sheet update plus Tuesday huddle reinforcement holds the line.
• 30-day deployment lifts diagnostic-to-quote pivot rate 18-24 points, Advisor sit-down close rate 6-12 points, and average ticket $320-$680 on the conversation. The economic ROI of the lesson on a 14-truck shop runs $180K-$420K of additional replacement revenue annually.