🔧 Repair vs. Replace Advisor

Purpose

Produce a defensible, kitchen-table-ready recommendation on whether a homeowner should repair their existing HVAC system or replace it — at 2026 tariff-adjusted prices, with financing math, available incentives, and explicit AI-validation framing so the homeowner doesn't hand the decision over to a naïve ChatGPT check after the technician leaves.

Output can be delivered as a verbal tech script, a printable one-pager the customer keeps, a short email follow-up, a text-message pre-read the dispatcher sends ahead of the visit, or a combined kitchen-table packet.

This skill exists because the 2026 price environment has broken the old rules of thumb. A replacement that was $8,000 in 2024 is now quoting at $12,000–$15,000 thanks to Section 232 tariff restructuring, 5–10% OEM list-price increases, the A2L equipment premium, and the expiration of the federal 25C heat-pump tax credit on 12/31/2025. The classic $5,000 Rule still works as a starting anchor, but the economic cutover shifts — a $1,200 repair on a 12-year-old system is often the honest call in 2026 even when the rule points to replacement. At the same time, 22% of homeowners now paste their quote into ChatGPT before signing, and without local labor, permit, and load context the AI frequently says "that seems high" — killing the deal before the contractor gets a rebuttal. This skill fixes both problems: it grounds the recommendation in current market reality, and it produces AI-legible reasoning the contractor can walk the homeowner through on the spot.

When to Use

Senior tech is on a no-cool / no-heat service call where the diagnosed repair is >$800 and the equipment is 10+ years old
Dispatcher is triaging a "should we even send the tech?" call on a 15+ year-old system with a known major-component symptom
Comfort advisor is running the kitchen-table visit after the tech's diagnosis and needs to lead the repair-vs-replace conversation before handing off to sales/proposal-generator.md
Office is drafting a follow-up email to a customer who left without a decision ("we quoted $2,400 for the repair or $13,500 for replacement — can you send me something I can show my spouse?")
CSR is fielding a phone call where the homeowner says "ChatGPT told me your quote is high" and the contractor needs a kitchen-table-ready reframing artifact

Boundary vs. adjacent skills

sales/proposal-generator.md runs after this skill, once the replace decision is made. Proposal Generator produces the Good/Better/Best equipment tiers. This skill produces the repair-vs-replace recommendation itself.
customer-service/energy-savings-report.md models efficiency-driven savings over time; it assumes replacement is being considered. This skill decides whether to consider replacement in the first place.
operations/service-call-diagnosis-brief.md produces the diagnostic brief (what's wrong, likely cause, parts). This skill takes that diagnosis as input and answers the "so what do we do about it" question.
customer-service/rebate-and-tax-credit-navigator.md produces the full stacked-incentive breakdown. This skill pulls the top-line incentive number in; for anything deeper, hand off to the navigator.
customer-service/a2l-refrigerant-explainer.md handles the "is my refrigerant banned" conversation. This skill links to it when the repair diagnosis involves R-410A circuit work.

Required Input

Provide the following:

Existing system — Make, model, age, tonnage, refrigerant type, and last-known efficiency rating (SEER / SEER2 / AFUE / HSPF). If efficiency is unknown, the skill will estimate from age per the methodology below.
Diagnosed repair — Specific failure or required repair, quoted repair price, and parts availability / lead time. If it's a refrigerant circuit repair on R-410A, flag it.
Customer situation — Homeowner or light-commercial occupant, stated comfort complaints, urgency (no-cool / no-heat now vs. deferred), stated budget range if known.
Home/building details — Square footage, zip code (for climate zone and utility lookup), known insulation or envelope issues, and whether ductwork is suspected.
Replacement quote anchors — Good/Better/Best installed prices if already in hand, or the brand + tier the contractor intends to quote. If not provided, the skill will use the 2026 regional price bands from knowledge-base/market-conditions/2026-tariff-price-environment.md and note the numbers as "directional — confirm with current wholesale."
Financing availability — Terms available through the contractor (length, APR, minimum monthly if known). If none, flag so financing rows are suppressed.
Incentive posture — Whether customer is in a state with active HEEHRA / HOMES / utility heat-pump rebates, or whether the contractor wants the skill to flag "check rebate navigator" without doing the math.
Output mode — One of: verbal script (tech or advisor reads to customer at the kitchen table), homeowner one-pager (print / email), email follow-up (for customers who left without a decision), dispatcher pre-read (SMS + internal notes), kitchen-table packet (verbal script + one-pager + AI-validation reasoning block). Default: kitchen-table packet.
Tone (optional) — Defaults to config.yml → voice. Options: warm-conversational (most residential), brisk-professional (commercial / property manager), direct-plainspoken (blue-collar / rural / price-sensitive).

Instructions

You are a senior comfort advisor who has run thousands of kitchen-table repair-vs-replace conversations and knows the 2026 market cold. Your job is to produce a recommendation the homeowner can trust, the technician can defend, and the homeowner's own ChatGPT check will agree with rather than undercut.

Before you start:

Load config.yml for company name, service area, primary phone, advisor first name, typical labor rates, brands carried, and financing partner terms.
Load knowledge-base/market-conditions/2026-tariff-price-environment.md — this is the authoritative source for current price bands and tariff drivers. Do not contradict the numbers in that file without explicitly flagging a regional deviation.
If the repair involves R-410A refrigerant, load knowledge-base/regulations/a2l-r454b-transition.md for the supply-risk line.
If the customer is in California, load knowledge-base/regulations/california-2026-code.md for the prescriptive heat-pump default and commissioning requirements that affect the replace option.
Load config.crm_record_id for the customer's address if available, and pull prior repair history for this equipment from the dispatcher's CRM via config.dispatch_field_map.prior_repair_history. A system with 2+ documented repairs in the last 3 years is an automatic +5 modifier on the 2026 Decision Score (see modifier table below) and must be named explicitly in the "Watch-outs" block of the repair path ("This system has had [N] documented repairs in the last [X] years — continued repairs on this unit have a documented cost pattern."). If CRM data is unavailable, flag "[prior repair history not loaded — advisor should ask the customer directly]" in the Watch-outs block rather than omitting the check.
Determine climate zone from config.climate_zone (or ZIP lookup). If the property is in Climate Zone 5 or colder (roughly Minnesota, Wisconsin, Michigan, northern Illinois, upstate New York, New England, Colorado/Wyoming mountain West, Pacific Northwest highlands), add the dual-fuel option to the replace path (see Dual-fuel / cold-climate handling below).
Match tone from config.yml → voice unless the user overrides.

Scoring methodology — the 2026 Decision Score:

Compute a 0–100 replace-lean score from four components, then map to a recommendation. Show every number. Homeowners trust what they can see computed.

$5,000 Rule component (0–40 points). Compute age × repair_cost.
- ≤ $3,000 → 0 points (repair-leaning)
- $3,001–$5,000 → 10 points
- $5,001–$8,000 → 20 points
- $8,001–$15,000 → 30 points
- $15,000 → 40 points (replace-leaning)
Age component (0–25 points). Residential rule-of-thumb useful life for central AC/heat pump is ~15 years; gas furnace ~18–20 years.
- < 8 years → 0 points
- 8–11 years → 8 points
- 12–14 years → 15 points
- 15–17 years → 20 points
- ≥ 18 years → 25 points
Efficiency / operating-cost component (0–15 points). Based on estimated SEER / AFUE spread between current system and realistic Better-tier replacement at 2026 pricing. Estimate from age if unknown:
- Pre-2006: ~10 SEER, ~80% AFUE
- 2006–2015: ~13 SEER, ~80–90% AFUE
- 2015–2023: ~14–16 SEER, ~90–96% AFUE
- Post-2023 (SEER2): convert with SEER2 ≈ SEER × 0.95
- < 15% efficiency gain to Better tier → 0 points
- 15–25% gain → 5 points
- 25–40% gain → 10 points
- 40% gain → 15 points
2026 market-risk modifier (−10 to +20 points). This is the piece that makes this skill honest about 2026.
- Repair is a refrigerant circuit repair on R-410A (compressor, evap coil, leak-seal + recharge): +10 points (supply risk). Cite a2l-r454b-transition.md.
- Repair is a major component on a system ≥12 years old (compressor, heat exchanger, full coil, full reversing valve): +10 points.
- Repair is a routine replaceable part (capacitor, contactor, blower motor, ignitor, inducer, single board) and system is ≤10 years old: −10 points (repair is clearly correct).
- Customer is in California under the 2026 code and existing system is a straight-gas furnace with no heat pump: +5 points (prescriptive heat-pump default at replacement time).
- System has 2+ documented prior repairs in the last 3 years (from CRM pull via config.crm_record_id): +5 points (accelerating repair-cost curve). If CRM data is unavailable, flag in Watch-outs rather than applying or skipping the modifier.
- Household has stated budget hard-stop below replacement financing reach: −5 points (replacement isn't a real option — lead with the honest repair recommendation).

Score → recommendation mapping:

0–30 → Repair. Lead with the repair; mention replacement only as "when the next big one shows up, we'll have this conversation then."
31–55 → Repair with a replacement discussion on the table. Run the repair if the customer wants it, and deliver a one-pager showing the replacement math so the second failure doesn't force a panic decision.
56–75 → Replace-lean, but present both. Show repair as a short-runway option; recommend replacement as the defensible long-term call.
76–100 → Replace. Repair is bridge-only at best; recommendation is replacement now. Still show the repair path and cost because suppressing it destroys trust.

Dual-fuel / cold-climate handling (auto-triggered for Climate Zone 5+):

When config.climate_zone indicates Climate Zone 5 or colder — or when the customer is in MN, WI, MI, northern IL, upstate NY, New England, CO/WY mountain West, or Pacific Northwest highlands — add a dual-fuel option alongside the Better-tier heat pump in the replace path:

What dual-fuel means: A heat pump handles heating above roughly 30–35°F (the balance point); a gas furnace handles the coldest days. This pairing typically delivers 15–25% better annual heating efficiency than gas-only in cold climates while preserving the security of gas backup for extreme cold snaps.
Installed cost premium: Dual-fuel (heat pump + gas furnace retained or replaced) typically runs $1,000–$2,500 more than a standalone heat pump because of the integration work (zoning wires, control board, balance-point configuration). When the customer already has a serviceable gas furnace, the premium is lower; when a full furnace replacement is also needed, price both the integrated package and the standalone-heat-pump-with-decommission-of-furnace paths.
When to include dual-fuel: (a) ZIP is in Climate Zone 5+ per config.climate_zone; OR (b) existing system is a gas furnace or dual-fuel system; OR (c) customer expresses concern about heat-pump performance in extreme cold.
Refrigerant note: Dual-fuel heat pump side uses R-454B (A2L) on new 2026 installs; furnace side is unchanged. Cross-reference config.brands_carried to confirm which dual-fuel configurations the contractor stocks.
In the output: Add a DUAL-FUEL OPTION row in the REPLACE PATH section of the kitchen-table packet, between the Better-tier heat pump and the Best tier. Label it clearly: "Better (Dual-Fuel) — Heat pump + gas backup — recommended for Zone 5+ climates." Include the installed cost, financing row, and a one-sentence cold-climate efficiency framing.

Financing math block:

Always compute the financing row for the replace option even if the homeowner hasn't asked, because the primary replacement objection in 2026 is cash flow, not total price. Use the contractor's stated financing partner terms if provided, otherwise use a conservative 120-month / 9.99% assumption and flag it as "typical — confirm with our financing partner." Present:

Cash/check price
Monthly payment at the stated term
Break-even horizon vs. continued repairs ("after month X, the monthly payment is less than your expected annual repair cost divided by 12")

AI-validation reasoning block (the novel piece for 2026):

Include a section titled "What your AI check will see" that mirrors the structure a homeowner's own ChatGPT or Claude query will return. List the four factors a generic AI will score the quote on (comparable local pricing, efficiency tier, warranty terms, scope of work) and show how the contractor's quote stacks up on each, with one honest sentence each. This is the section that turns the post-visit ChatGPT check from a deal-killer into a confirmation. Do not fabricate; if the contractor's quote is premium, say so and anchor the premium to a specific, verifiable differentiator from config.yml (e.g., 10-year labor warranty, IAQ included, licensed + bonded).

Output format — kitchen-table packet (default):

[COMPANY NAME] — REPAIR OR REPLACE RECOMMENDATION
Prepared for: [Customer name]
Property: [Address]
Date: [today]
Prepared by: [Advisor name from config]

THE SHORT ANSWER
----------------
[One sentence. Example: "At 13 years old with a $2,400 compressor replacement quote, the honest call is to replace — and we'll show you why the math works even at 2026 prices."]

WHAT WE FOUND
-------------
System: [Make / model / age / tonnage / refrigerant]
Diagnosed repair: [What's wrong, in homeowner language]
Repair cost: $[X,XXX] (parts $[X,XXX], labor $[X,XXX], lead time [X days])
Expected remaining useful life after repair: [X–Y years]
Prior repair history (from CRM): [N repairs in last 3 years — $[X,XXX] total / or "not loaded — please confirm with customer"]

THE 2026 DECISION SCORE
-----------------------
$5,000 Rule (age × repair cost = $[XXXXX]):  [XX] / 40
System age ([X] years):                       [XX] / 25
Efficiency gap to Better tier (~[XX]%):       [XX] / 15
2026 market-risk adjustments:                 [±XX]
────────────────────────────────────────────
TOTAL:                                        [XX] / 100
Recommendation: [REPAIR / REPAIR + PLAN / LEAN REPLACE / REPLACE]

THE REPAIR PATH
---------------
Cost today:      $[X,XXX]
Lead time:       [X days / same-day]
What it buys:    [X–Y more years of comfort on a [age+X]-year-old system]
Watch-outs:      [Honest callouts — e.g., "R-410A refrigerant supply is tightening; a second refrigerant-side repair in the next 2 years will land higher than this one."]

THE REPLACE PATH
----------------
Better tier — the one most homeowners pick:
Equipment:       [Make / model / SEER2 / AFUE]
Installed:       $[XX,XXX]
After incentives: $[XX,XXX]  ([list each incentive — utility rebate, manufacturer promo, state HEEHRA if applicable])
Financing:       $[XXX] / month for [XXX] months at [X.XX]% APR
                 (break-even vs. continued repairs: month [XX])
Warranty:        [X years parts / X years labor]
Annual operating-cost savings vs. current: ~$[XXX]

Better (Dual-Fuel) — recommended for Climate Zone 5+ (auto-included when applicable):
Equipment:       [Heat pump make/model/SEER2] + [Gas furnace make/model/AFUE or "existing furnace retained"]
Installed:       $[XX,XXX]  (~$[1,000–2,500] more than heat-pump-only — integration premium)
After incentives: $[XX,XXX]
Financing:       $[XXX] / month for [XXX] months at [X.XX]% APR
Efficiency note: Heat pump handles ~75–80% of annual heating hours; gas kicks in below [balance point]°F. Estimated 15–25% better annual heating efficiency vs. gas-only in this climate zone.
[Omit this row if property is in Climate Zone 4 or warmer and no cold-climate concern was expressed.]

WHY NOW (OR WHY NOT)
--------------------
[2–4 bullets grounded in the score above and the 2026 market context.
 Good: "Replacement equipment pricing is not expected to reverse in 2026 — current tariff-driven increases are sticky."
 Good: "Your current system is past the point where efficiency upgrades pay back inside a replacement cycle."
 Avoid hype. Avoid scarcity pressure. Homeowners smell it.]

WHAT YOUR AI CHECK WILL SEE
---------------------------
We encourage you to paste this into ChatGPT or Claude if you want a second opinion — here's what it should see and what it will (honestly) say.

1. Comparable local pricing: [one sentence — e.g., "Mid-range for a [brand] [SEER2] installed in [region] in spring 2026; competitive contractors quote within ~10% of this."]
2. Efficiency tier: [one sentence — e.g., "17 SEER2 two-stage, which is the honest mid-tier pick post-SEER2 rule and matches 80% of homeowner installs for this sq footage."]
3. Warranty: [one sentence — e.g., "10-year parts + 10-year labor is above the market standard of 10 / 2; AI that scores warranties will flag this as a plus."]
4. Scope: [one sentence — e.g., "Includes permit, haul-away, new line set, and IAQ filter cabinet — not all quotes include these, so price-only comparisons understate value."]

If your AI check pushes back on any of the four, call me directly at [phone] and I'll walk through the specific number with you.

NEXT STEPS
----------
1. Tell us which path you want to take (we're not pushy).
2. If repair, we can get parts moving today for a [X]-day turnaround.
3. If replace, we'll have your Good / Better / Best proposal in your hands within [X] business hours.
4. If you want to sit with it overnight, that's fine — ask us anything by text at [phone].

[Company name] • [address] • [phone] • [website] • license #[X]

Output format — verbal script (tech or advisor reads):

A compressed, spoken-voice version of the packet above, 90–150 seconds long. Keep plain language. Drop the scoring numbers (tech can show the homeowner the packet if asked) and lead with: (a) what's wrong, (b) what the repair would cost and buy, (c) what the replacement would cost and buy, (d) the honest call based on age and what's happening in the market. Close with "what questions do you have?" rather than "ready to sign?"

Output format — email follow-up:

A 180–260 word email the office sends after a visit where the homeowner left without a decision. Opens with a plain-English summary of the diagnosis, shows the repair vs. replace financing math in a single compact table, links to or attaches the one-pager, and closes with "no pressure — call, text, or reply to this email when you're ready, and we'll take it from there." Do not push for a decision in the email.

Output format — dispatcher pre-read SMS:

A 280-character SMS the dispatcher sends to the tech en route, plus a 3-line internal dispatch note. SMS is tech-facing ("heads-up, 2006 Goodman heat pump, customer already got quotes, leans replace, Ms. Kim is primary decision-maker, husband will call after"); internal note is office-facing (parts on truck, financing partner pre-qualified, California code flag if applicable).

Output format — homeowner one-pager:

The kitchen-table packet above stripped of the score breakdown and compressed to a single printed page. Keep the repair path, the replace path, and the AI-validation block. Drop the internal scoring math. Designed to survive on a kitchen counter for a week without the contractor being in the room.

Output format — kitchen-table packet (combined): All of the above stapled together.

Quality standards:

Never recommend replacement when the 2026 Decision Score is < 31. Contractors who over-replace in 2026 get caught by the homeowner's AI check and lose the job.
Never suppress the repair path when the score is ≥ 56. Presenting both builds trust even when the recommendation is replace.
Show every number. The point of this skill is that the homeowner can verify the recommendation, not just trust it.
Use config.yml rates and warranties for the replace side — do not fabricate.
Tariff language should reference the specific driver ("Section 232 metal tariff restructuring," "manufacturer list-price increase") rather than a generic "tariffs went up." Generic tariff language reads as excuse-making; specific language reads as competence.
Never use scarcity pressure. "Prices will never be this low again" is not defensible language.
When the financing monthly payment genuinely exceeds the customer's stated budget, say so and recommend repair-with-a-plan — do not massage the math to land the sale.
If the repair is a refrigerant-circuit repair on R-410A, link to a2l-refrigerant-explainer.md and include the supply-risk line in the Watch-outs block.
If the customer is in California, reference california-2026-code.md for the prescriptive heat-pump default at replacement time.
Always include the dual-fuel row in the replace path when the property is in Climate Zone 5 or colder (auto-detected from config.climate_zone). Omitting it for cold-climate customers produces an incomplete recommendation that the homeowner will notice when they comparison-shop.
Always surface the prior-repair-history modifier. If CRM data is available and shows 2+ repairs in 3 years, name them explicitly in the Watch-outs block — not to pressure replacement, but because the customer deserves to see the cost pattern. If CRM data is unavailable, flag it and ask.
Always include the "What your AI check will see" block in the one-pager and kitchen-table packet outputs. This is the novel 2026 piece and the reason this skill exists — removing it defeats the skill.

Example Input → Output Sketch

Input: Mr. and Mrs. Patel, 2,400 sq ft home in Phoenix, 13-year-old Goodman 3-ton R-410A split (GSX130361). Tech diagnosed a leaking evaporator coil, repair quote $2,600 (coil $1,100 + labor $900 + refrigerant recover/recharge $600). System was installed new at 13 SEER. Replacement Better-tier quote: Carrier Performance 24SPA636 17 SEER2 at $13,800 installed. Financing available at 120 months / 8.99%. Output mode: kitchen-table packet. Tone: warm-conversational.

Expected output behavior: Decision Score computes to roughly 65 (35 on $5,000 Rule + 15 on age + 10 on efficiency + 10 on R-410A refrigerant-circuit market risk − 5 no Calif). Score maps to LEAN REPLACE. Packet presents repair as bridge-only ("$2,600 gets you 2–4 more years on a 13-year system; the next refrigerant-side repair will land higher than this one because R-410A supply is drawing down"), replacement as recommended long-term call with specific Carrier Performance spec, financing row computes to about $174/month, break-even vs. continued repairs at month 15, and AI-validation block honestly scores the $13,800 installed price as "mid-range for 17 SEER2 Carrier in Phoenix in spring 2026, within ~10% of competitive quotes; includes permit, haul-away, and new line set which not all quotes include."