ChatGPT for Service Advisor Scripts: Close More Approvals
ChatGPT writes solid service advisor scripts when you give it the actual measurements, pricing, and vehicle specifics. Generic prompts produce generic scripts. The 9 prompts below include the context fields — pad thickness, fluid condition, exact pricing — that make the difference between a script a customer believes and one they tune out.
Service advising is communication work. A technician can find the right problem and recommend the right fix, but if the advisor can’t explain it clearly on the phone or in a walk-around, the customer declines the repair and drives away with a car that isn’t safe. That scenario costs the shop revenue and puts the customer at risk — neither outcome is acceptable.
ChatGPT is genuinely useful for service advisors when you use it for the right tasks. It writes first drafts fast. It handles the repetitive scenarios — declined repairs, estimate follow-ups, deferred maintenance reminders — that take up disproportionate time in a busy write-up bay. Where it falls short is anything requiring real-time vehicle information or genuine mechanical judgment. Use it for communication scaffolding, not diagnostic decisions.
The prompts below are built around a fictional shop called Trident Automotive in Columbus, Ohio. Replace the shop name, advisor name, pricing, and vehicle specifics. The structure of each prompt is the part worth keeping.
The 9 Prompts
Prompt 1: Inbound Call Script for a New Customer
When to use it: A first-time customer calls to schedule an appointment. You want to gather the right information without a 12-minute intake that frustrates the caller.
The prompt:
Write a phone script for a service advisor at Trident Automotive in Columbus, Ohio. A new customer is calling to schedule an appointment for a brake noise complaint on their 2019 Honda CR-V.
The script should:
- Greet the customer warmly using the advisor's name (use [ADVISOR NAME] as placeholder)
- Ask for the customer's name, phone number, and mileage
- Ask one open-ended question about when and how the noise occurs (stopping from highway speed vs. city driving, metal-on-metal vs. grinding)
- Offer two appointment times: tomorrow morning at 8am or Thursday at 1pm
- Confirm what was said and set expectations: "Our tech will inspect the brakes and we'll call you with an estimate before we start any work."
Tone: Confident and efficient. This is a professional shop, not a dealership conveyor belt.
Do not use "absolutely," "no problem," or "awesome."
Length: Under 200 words as spoken dialogue.
What the output looks like: ChatGPT produces a clean two-minute script with a clear opening, a single diagnostic question about noise characteristics, and a firm scheduling close. The instruction to avoid filler words matters — without it, the output defaults to the call-center phrases that signal a scripted interaction.
Customization tip: Add your actual appointment windows and whether you use a specific scheduling system like Shop-Ware or Mitchell 1. If your shop texts appointment confirmations automatically, include that in the script close.
Grade: Ready to use with light editing.
Prompt 2: Estimate Presentation Phone Call
When to use it: You have the diagnosis and estimate ready. You need to call the customer and present $1,400 in repairs clearly without overwhelming them.
The prompt:
Write a phone script for a service advisor presenting a repair estimate to a customer over the phone. The vehicle is a 2017 Ford F-150 with 87,000 miles. The customer came in for an oil change, and we found three additional issues.
Issues found:
1. Front brake pads worn to 2mm, rotors worn beyond minimum thickness — $520 parts and labor
2. Serpentine belt cracked with visible fraying — $195 parts and labor
3. Cabin air filter clogged (can wait) — $45
The script should:
- Lead with what the customer came in for (oil change is done and costs $89)
- Present the brake job first as safety-critical, with the specific measurements
- Present the belt second as a failure risk in the next 10,000 miles, not an emergency today
- Present the cabin filter as optional with a clear note it can be deferred
- Offer to approve items individually or all at once
- Confirm the total if they approve everything: $849
Tone: Informational and direct. The customer is an adult who can make their own decisions.
Do not say "I just wanted to reach out" or "I have some concerns."
What the output looks like: The output sequences the repairs correctly — safety items first, with specific measurements rather than vague recommendations, then the higher-risk-but-not-critical belt, then the optional filter. The individual approval option reduces the all-or-nothing pressure that causes customers to decline everything.
Common mistake: Advisors run this prompt without including the actual measurements (2mm pad thickness, visible fraying). Without those specifics, ChatGPT writes vague output like “your brakes are getting low.” The measurements are what make the recommendation credible.
Grade: Strong. Requires your actual measurements and pricing.
Prompt 3: Declined Repair Follow-Up Text
When to use it: A customer declined a brake job or other safety-critical repair at the time of service. You want to follow up without being pushy — and create a paper trail that the customer was informed.
The prompt:
Write a follow-up text message from Trident Automotive to a customer who declined a brake job at today's visit. The vehicle is a 2016 Toyota Camry with 74,000 miles. The front pads measured 2mm and the rear pads measured 3mm.
The text should:
- Reference the specific measurements from today's inspection
- Note that 2mm is at the service limit for most manufacturers
- Remind them that stopping distance increases significantly below 2mm
- Invite them to call or text to schedule when they're ready
- Include the shop phone number: (614) 555-0271
Keep it under 100 words. Do not use guilt-based language. Do not start with "Unfortunately."
This message also serves as documentation that the customer was informed of the finding.
What the output looks like: A short, factual message that references the specific measurements, explains the safety threshold plainly, and ends with a clear call to action. The documentation framing is important for liability purposes — this text is evidence that the customer was told what was found and when.
Grade: Ready to use. One of the highest-value prompts in this list.
Prompt 4: Upsell Conversation for a Deferred Maintenance Item
When to use it: During a write-up, you want to introduce a deferred maintenance item — transmission fluid, coolant flush, differential service — without it feeling like an add-on sales pitch.
The prompt:
Write a service walk-around script for a service advisor at Trident Automotive introducing a transmission fluid service recommendation. The vehicle is a 2015 Jeep Grand Cherokee with 68,000 miles. The manufacturer recommends transmission fluid change at 60,000 miles. The customer came in for a routine oil change.
The script should:
- Reference the mileage and the manufacturer's service interval specifically
- Explain what happens to automatic transmission fluid over time (heat breakdown, friction modifier depletion)
- Note that the fluid is dark brown on inspection (use this as the specific finding)
- Give the price: $189 for a transmission fluid drain and refill
- Offer to add it to today's service or schedule it separately
Avoid: "We recommend," "you should consider," "it might be a good idea." These are weak. State what the maintenance interval is and what the fluid condition shows.
Tone: A technician explaining a finding, not a salesperson pitching an add-on.
What the output looks like: The output leads with the mileage-versus-interval gap (68,000 vs. recommended 60,000) and the dark brown fluid as a physical finding. It explains why heat breaks down friction modifiers in plain terms and offers the two-path close. The instruction to avoid weak hedging language produces noticeably stronger output.
Customization tip: This structure works for any fluid service. Swap the fluid type and the manufacturer’s interval. For coolant, add the pH test strip result. For differential fluid, add the metal particle finding if applicable.
Grade: Strong. The vehicle specifics are what separate this from a generic upsell pitch.
Prompt 5: Estimate Approval Email for a Complex Repair
When to use it: A customer is out of town or prefers email. You need to present a multi-item estimate clearly enough that they can approve it without a phone call.
The prompt:
Write an estimate approval email for Trident Automotive. The customer's 2013 BMW 328i has been diagnosed with the following:
1. Valve cover gasket leak — oil seeping onto exhaust, minor smoke on startup — $485 parts and labor
2. Front control arm bushings worn, causing pulling and uneven tire wear — $620 parts and labor
3. Brake fluid flush due at 50,000 miles (vehicle is at 54,000) — $95
The email should:
- Summarize what was found and why each item matters
- Clearly separate the urgent repairs (valve cover leak near exhaust) from the maintenance item
- List each item with its price and a checkbox option to approve individually
- Give a total for all three: $1,200
- Explain the shop's approval process (reply to this email or call the shop number)
- Include the shop number: (614) 555-0271
Do not use "comprehensive inspection" or "state-of-the-art." Do not describe the shop in marketing terms.
Length: Under 400 words.
What the output looks like: A clean, structured email with itemized findings, clear priority indicators, and individual checkboxes. The instruction to separate urgency from maintenance helps customers feel they have agency rather than being pressured into approving everything at once. The BMW-specific framing (valve cover to exhaust is a fire risk on some platforms) gives the safety explanation credibility.
Grade: Ready to use with your actual findings and pricing.
Prompt 6: Warranty Concern Response
When to use it: A customer is upset because a repair you did three months ago is having issues again. You need to respond professionally without admitting liability before you’ve diagnosed the vehicle.
The prompt:
Write a response script for a service advisor at Trident Automotive handling a warranty concern call. A customer is calling because their vehicle is making a similar noise to the one they brought it in for 90 days ago. The original repair was a front wheel bearing replacement at $380.
The script should:
- Acknowledge that a return issue is frustrating without admitting the original repair failed
- Ask clarifying questions: Is the noise the same side? Same road speed range? Same conditions?
- Invite them to bring the vehicle in for a no-charge inspection
- Explain the shop's 12-month/12,000-mile parts and labor warranty
- Avoid defensive language or implying the customer caused the problem
Tone: Calm and professional. The goal is to get the vehicle back into the shop so it can be properly diagnosed, not to win an argument on the phone.
What the output looks like: The output walks through the diagnostic questions calmly, references the warranty clearly, and closes with a specific invitation to bring the vehicle in rather than a vague “we’ll take a look at it.” The instruction to avoid defensive language is the key constraint here — ChatGPT will default to defensiveness if you don’t tell it not to.
Grade: Ready to use. This is one of the harder calls to handle, and the script structure holds up.
Prompt 7: Deferred Repair Reminder Email (30-Day Follow-Up)
When to use it: A customer deferred a repair 30 days ago. You want to follow up with a short, specific reminder that references what they declined and why it matters.
The prompt:
Write a 30-day follow-up email from Trident Automotive to a customer who deferred a serpentine belt replacement 30 days ago. The belt showed cracking and fraying at the last visit. The vehicle is a 2014 Chevrolet Equinox with 91,000 miles.
The email should:
- Reference the specific finding from the previous visit (cracking and fraying on the serpentine belt)
- Explain briefly why a belt failure matters: stranded vehicle, potential damage to accessories driven by the belt
- Note that the price is still $195 (no price increase since the last visit)
- Include a direct link or call-to-action to schedule: "Call us at (614) 555-0271 or reply to this email."
Length: Under 150 words.
Do not start with "I hope this email finds you well."
Do not use "just checking in" or "circling back."
What the output looks like: A short, direct email that references the specific finding, explains the consequence of ignoring it in practical terms (stranded vehicle rather than vague “damage”), and makes scheduling easy. The constraints on opener phrases prevent the AI-sounding phrases that signal a mass mailer rather than a personal message.
Grade: Ready to use.
Prompt 8: Multi-Vehicle Fleet Customer Summary
When to use it: You service a local fleet account — a plumber, a landscaping company, a contractor — with multiple vehicles. You need to send a monthly summary of vehicle statuses and upcoming needs.
The prompt:
Write a monthly vehicle status summary email for a fleet customer account at Trident Automotive. The customer is Blue Ridge Landscaping, operating 4 vehicles. Summary for January 2026:
Vehicle 1: 2019 Ford F-150 (Unit 1) — Oil change completed 1/8, next due 3/8. Rear brake pads at 4mm — monitor at next service.
Vehicle 2: 2020 Ram 1500 (Unit 2) — Oil change completed 1/15, next due 3/15. No issues noted.
Vehicle 3: 2018 Chevy Colorado (Unit 3) — Overdue for oil change by 1,200 miles. Left rear tire at 3/32" tread — needs replacement before next route.
Vehicle 4: 2021 Ford Transit (Unit 4) — Oil change completed 1/20, next due 3/20. Transmission fluid due at next service (approaching 60,000 miles).
The email should:
- Summarize each vehicle's status clearly
- Flag Unit 3 as needing immediate attention (tire)
- List upcoming due items across the fleet for the next 60 days
- Close with a direct contact for scheduling
Tone: Business correspondence. This is a B2B email, not a consumer-facing message.
What the output looks like: A well-structured fleet summary with clear status flags, an immediate-action callout for the tire, and a forward-looking maintenance schedule. Fleet customers respond well to this format because it treats them as business operators managing costs, not individual consumers being sold services.
Grade: Strong. The more complete your vehicle status data, the more useful the output.
Prompt 9: Response to a Negative Google Review
When to use it: A customer left a 1- or 2-star Google review about their service experience. You need a professional response that addresses their concern without escalating publicly.
The prompt:
Write a response to a Google review for Trident Automotive. The customer left a 2-star review saying: "Brought my car in for an oil change and they told me I needed $1,200 in repairs. Felt like a lot of upselling. Won't be back."
The response should:
- Thank them for taking the time to leave feedback (genuine, not sycophantic)
- Acknowledge that a $1,200 estimate is significant and can feel overwhelming
- Explain briefly that the shop provides itemized estimates and customers are never required to approve additional work
- Invite them to call the owner directly to discuss their specific visit: (614) 555-0271
- Stay under 100 words
- Not be defensive or imply the customer is wrong
Tone: Professional and grounded. This response will be read by future customers, not just the reviewer.
What the output looks like: A short, professional reply that acknowledges the customer’s reaction without arguing with them. The phone number invitation is critical — it moves the conversation off the public forum while signaling to other readers that the shop is responsive. ChatGPT handles this prompt consistently well when you specify the word limit.
Grade: Ready to use with minimal editing.
Common Prompt Mistakes in Auto Repair Service Advising
Not giving ChatGPT the actual measurements. The difference between “your brakes are low” and “your front pads measured 2mm, which is at the manufacturer’s minimum” is the difference between a customer declining a repair and a customer approving it. Always put the actual numbers in the prompt.
Letting ChatGPT invent pricing. If your prompt doesn’t include the repair cost, ChatGPT will either omit pricing or invent a number. Always include your actual price. A script with the wrong price is worse than no script.
Skipping the tone instruction. Every prompt in this list includes a tone instruction. Without it, ChatGPT defaults to a generic corporate register that sounds like a national chain rather than a local shop. Write one sentence about who the advisor is and who the customer is.
Using the output without reading it aloud. Approval scripts need to be spoken, not read. Have your advisors read every script out loud before using it. ChatGPT writes formally by default, and formal phrasing that looks fine on screen sounds stiff on a phone call. You will need to loosen the language.
Forgetting to remove the placeholders. Prompts in this list use bracketed placeholders like [ADVISOR NAME] and [RENEWAL LINK]. These are easy to miss. Set up a simple find-and-replace in whatever system you use to store scripts.
How to Customize These Prompts for Your Shop
The variables that change the output quality the most:
- Shop name and city — local specificity always outperforms generic phrasing
- Advisor name — first-name personalization reads differently than “a service advisor”
- Actual measurements — pad thickness, fluid color, mileage gap from service interval
- Your pricing — real numbers are the difference between a usable script and a template
- Your warranty terms — if you offer a 12-month/12,000-mile warranty, say so in the relevant prompts
- Your shop management system — if you use Mitchell 1 or Shop-Ware for texting, the call-to-action changes
One habit that consistently improves output quality: include one constraint about what NOT to say. “Do not use ‘I just wanted to reach out’” removes one of the most common AI tells. “Do not start with ‘I hope this email finds you well’” eliminates another. Two constraints like these, added to any prompt, noticeably shift the output away from sounding like a chatbot and toward sounding like a specific person at a specific shop.
The goal is not to remove the advisor from the communication. It’s to give the advisor a starting point that takes 30 seconds to generate and 2 minutes to edit, rather than starting from a blank screen when they have six cars to write up and a customer waiting at the counter.