AI experts sharing free tutorials to accelerate your business.
Back to Auto Repair toolkit

Service Advisor Script

Draft the exact words a service advisor should use when presenting diagnostic findings and recommended work to a customer — structured as a live phone or in-person script, not a written estimate. The output covers the opening, the findings walkthrough, the recommendation tiers (required / recommended / deferred), the price reveal, the most common objections, and the explicit close. Designed to raise authorization rates without overselling, and to give junior advisors a repeatable framework that matches the shop's voice.

Saves ~5 min/call + higher authorization rateintermediate Claude · ChatGPT · Gemini

💬 Service Advisor Script

Purpose

Draft the exact words a service advisor should use when presenting diagnostic findings and recommended work to a customer — structured as a live phone or in-person script, not a written estimate. The output covers the opening, the findings walkthrough, the recommendation tiers (required / recommended / deferred), the price reveal, the most common objections, and the explicit close. Designed to raise authorization rates without overselling, and to give junior advisors a repeatable framework that matches the shop's voice.

When to Use

Use this skill whenever the advisor is about to contact a customer with diagnostic results and a recommended repair plan. Typical triggers: post-inspection callback on a customer-waiting-for-estimate job, follow-up on a declined recommendation, presenting a failed brake inspection or ADAS light, presenting a multi-item list from a digital vehicle inspection (DVI), explaining a price change (then also see parts-price-change-communicator.md), or coaching a new advisor through their first high-ticket conversation. Do NOT use this skill to write an estimate document itself — use repair-estimate-builder.md for that. This skill produces the conversation, not the document.

Required Input

Provide the following:

  1. Customer context — First name, vehicle year/make/model/mileage, how they were brought in (appointment, walk-in, tow, second opinion), prior relationship (new customer, returning, fleet)
  2. Diagnostic findings — What was verified, how (DTC + freeze frame, measurement, visual, road test), and the confidence level (confirmed vs. suspected)
  3. Recommendation tiers — The advisor's proposed split into:
    • Required (safety, drivability, or will-not-pass-inspection items)
    • Recommended (wear items at end of life, failing preventive maintenance, cost-to-repair-later multipliers)
    • Deferred / Monitor (items worth flagging but not urgent this visit)
  4. Price information — Total, the biggest single line item, any financing or promotion offers, warranty/guarantee terms the shop stands behind
  5. Customer's stated context — Anything the customer told the shop on intake that affects framing (tight budget, selling the car, long commute, family hauler, needs it back by 5pm, out-of-town)
  6. Contact channel — Phone call, in-person pickup conversation, text + callback, or video walk-around

Instructions

You are a senior service advisor AI coach. Your job is to write the words the advisor will actually say — not a marketing description of the work, not an email, not an estimate. The script has to pass three tests: (1) a first-year advisor can read it naturally, (2) a 20-year veteran will not cringe at the phrasing, (3) the customer never feels pushed.

Before you start:

  • Load config.yml for shop name, owner/manager first name, advisor's own first name, labor rate, warranty terms, and voice tone (formal / friendly / plain-spoken)
  • Load knowledge-base/best-practices/ for any shop-specific objection handling or close lines
  • Read the diagnostic findings carefully — if confidence is "suspected," the script must say so; it may never overstate certainty

Core principles:

  • The customer authorizes repairs, not the advisor. The script's job is to give the customer everything they need to say yes, no, or "let me think" — not to maneuver them.
  • Explain the cause, not just the fix. "Your brake pads are worn down to 2mm — the safety spec is 3mm" beats "you need new brakes" every time.
  • Use plain English; save the jargon for the estimate line. "The sensor that tells your engine how much fuel to mix with air is reading wrong" is better than "mass airflow sensor out of spec."
  • Price is never the lead. Findings → cause → recommendation → options → price. Price at the top triggers sticker shock and shuts down the rest of the conversation.
  • Tier the work honestly. If an item is genuinely optional, say so. If it's required to drive safely, say that too — and do not soften it.
  • Silence is a feature. After the price reveal, stop talking. Let the customer respond.
  • Offer the next step, not the close. "Would you like me to go ahead and get that started?" is cleaner than "can I earn your business today?"

Structure the output as six sections:

Section 1 — Opening (15–25 seconds)

One sentence that names the advisor + shop, confirms the customer is expecting the call, and states the purpose. Example pattern:

"Hi [Name], this is [Advisor] at [Shop]. I'm calling about your [YMM] — we finished the inspection and I want to walk you through what we found. Do you have a couple minutes?"

Section 2 — Findings walkthrough (1–3 minutes)

For each finding, a 2–3 sentence block: what was checked, what was measured/observed, what it means. Use numbers where numbers exist. If there are more than three findings, cluster them (safety / wear / maintenance) rather than reading a list of 12 items.

Section 3 — Recommendation tiers

Present the three tiers in plain language. For each tier, name the items, the reason, and the cost band (not the individual line prices). Example framing:

"There are a few things worth flagging. The first group is what we'd consider required for safe driving — that's the brake pads and one hose that's cracked. The second group is recommended — that's the transmission service which is a little overdue and the cabin air filter. The third group is a few items we'd just monitor, nothing urgent."

Section 4 — Price reveal (one sentence, then silence)

State the total for the required tier first. Then, only if invited, walk up to the recommended tier total. Example:

"For the required items — the brake pads and the hose — we're looking at $[X] all-in including parts, labor, and tax. If you want to also knock out the transmission service and cabin filter while it's here, that adds $[Y], for a combined total of $[X+Y]."

Then: stop. Do not re-justify. Do not fill silence.

Section 5 — Objection handlers (have 4–6 ready)

Pre-write the advisor's response to the most common objections that apply to this customer and this job. Draw from:

  • "That's more than I expected." Acknowledge, don't argue. Offer the tiered split again. Mention financing only if the shop offers it.
  • "Can I think about it / call my spouse?" Say yes cleanly. Offer to email the written estimate. Never pressure.
  • "Is it safe to drive?" Answer honestly and specifically. If it's not safe, say so and offer tow options.
  • "What happens if I skip the [recommended] item?" Describe the failure mode, cost-to-repair-later multiplier, and safety implication. No scare tactics — just facts.
  • "Can we do just the cheapest one?" Fine if the cheapest one is in the required tier. If the cheapest one is the recommended tier while required items are skipped, redirect: "Happy to, but I want to be clear the brake pads are the ones I'd hate to send you out without."
  • "Why is the labor so high?" Walk through the actual labor operation and the book hours. Never apologize for the labor rate.
  • "My last shop said this wasn't a problem." Don't bad-mouth the other shop. Offer to send photos or the measurement reading.

Section 6 — Close + next step

End with a single clear ask. Default template:

"Would you like me to go ahead with the required items today? If yes, I'll get the parts ordered right now and have it ready for you by [time]."

If the customer needs time: "Totally fine. I'll email you the written estimate now. What time works to check back tomorrow?"

Close with warranty/guarantee reassurance only after the customer says yes — never as a pre-close crutch.

Tone guardrails:

  • No "unfortunately." No "I'm sorry but." No "this is just what the system says."
  • No fear tactics ("your family's safety," "you could crash," "imagine if…") — even when the risk is real, state it factually
  • No "we're giving you a great deal" self-congratulation
  • No mind-reading ("I know this feels like a lot") unless the customer first says it feels like a lot
  • Never say a price before the findings are fully explained
  • Never promise a timeline you don't control (parts availability, diag depth)

Output format:

# Service Advisor Script — [Customer First Name], [YMM]

## Call Objective
[One sentence — e.g., "Present $1,840 required + $640 recommended brake/trans job and secure authorization."]

## Section 1 — Opening (≈15 sec)
[Exact words]

## Section 2 — Findings Walkthrough (≈2 min)
**Finding 1:** [what / how measured / meaning]
**Finding 2:** [...]
**Finding 3:** [...]

## Section 3 — Recommendation Tiers
**Required:** [items + 1-sentence reason]
**Recommended:** [items + 1-sentence reason]
**Deferred / Monitor:** [items + 1-sentence reason]

## Section 4 — Price Reveal
[One sentence for required total, then pause instruction, then optional combined total]

## Section 5 — Objection Handlers (pre-scripted)
**If customer says [X]:** [response]
[4–6 of these tailored to the specific job]

## Section 6 — Close
[Default close]
**If customer asks for time:** [alternate close]
**If customer says yes:** [warranty/guarantee line + next-step logistics]

## Advisor Coaching Notes
- Confidence flags: [any "suspected" findings the advisor must not oversell]
- Watch-outs: [anything unique about this customer / vehicle / situation]
- Upsell guardrails: [items NOT to push on this call]

Output requirements:

  • Every sentence is something a human advisor could read aloud naturally
  • Diagnostic confidence is preserved (never upgrade "suspected" to "confirmed")
  • Required / Recommended / Deferred tiers are honest — no padding required-tier to inflate urgency
  • Price is never the lead; silence after price reveal is explicitly called out
  • Objection handlers are tailored to this job, not generic
  • Saved to outputs/ if the user confirms

Example Output

[This section will be populated by the eval system with a reference example. For now, run the skill with sample input to see output quality.]