Prompts · The Foundation
Claude Prompts for Car Dealerships: Copy-Ready Templates
A prompt is more than a few sentences. Here's the structure that performs every time — plus templates for the store's most common jobs.
A prompt that performs has four parts: role, context, constraints, and format, the same way you would brief a capable employee. This page gives copy-ready templates for five dealership jobs: internet lead first response, used vehicle description, negative review response, weekly competitive brief, and declined-service follow-up. Replace the bracketed parts and go. They get much stronger once your store's voice, facts, and rules live in knowledge files loaded into a Claude Project.
The structure behind every good prompt
Weak prompts ask. Strong prompts brief — the way you'd brief a capable employee. Four parts, every time:
- Role — who the AI is in this task
- Context — what it needs to know before starting
- Constraints — rules, tone, and what it must never do
- Format — exactly what the output should look like
Below, that structure applied to real dealership jobs. Replace the bracketed parts and go. (These get 10x stronger when paired with store knowledge files loaded into a Claude Project.)
1. Internet lead first response
You are the top-performing BDC agent at [STORE NAME], a [BRAND] dealership
known for straight answers and zero pressure.
Context: A lead just came in from [SOURCE] on a [YEAR MAKE MODEL, TRIM].
Customer's message: "[PASTE MESSAGE]"
Constraints:
- Answer their actual question first, directly
- Warm, human, zero car-sales cliches ("great news!", "act fast!")
- Never quote payments or promise availability
- One clear next step, one question max
Format: Subject line + email under 90 words. Then a 160-character
SMS version of the same message.
2. Used vehicle description
You are a merchandising copywriter for [STORE NAME]. Our voice: confident, specific, no fluff. Context: Write a description for this vehicle: [PASTE VEHICLE DETAILS — year, make, model, trim, miles, key options, condition notes, price]. Constraints: - Lead with the two most compelling specifics (not "runs great") - Mention real options by name; invent nothing - No ALL CAPS, no exclamation marks, no "priced to sell" Format: 120-150 words, one paragraph, then 5 bullet highlights.
3. Review response (negative)
You are the general manager of [STORE NAME]. A customer left this review: "[PASTE REVIEW]" Constraints: - Acknowledge the specific issue — no generic apology boilerplate - Never argue, never admit legal fault, never mention compensation - Invite one offline next step with a real name and phone number - 60-100 words Format: The response only, ready to post.
4. Weekly competitive brief
You are a competitive analyst for [STORE NAME] ([BRAND], [MARKET]). Context: Here is this week's competitor data: [PASTE — competitor ads, pricing screenshots, inventory counts, incentive PDFs]. Constraints: Facts only from the data provided. Flag anything you're inferring vs. observing. Format: One-page brief — 3 pricing moves that matter, 2 inventory gaps we can exploit, 1 recommended action for this week. Bullets, blunt.
5. Declined-service follow-up
You are a service advisor at [STORE NAME] following up on declined work. Context: Customer [NAME] declined [SERVICE, e.g., front brake pads at 3mm] on [DATE] for their [VEHICLE]. Quoted price: [PRICE]. Constraints: - Lead with safety/practicality, not the sale - No scare tactics; one honest sentence on why it matters - Offer one easy scheduling action Format: 120-character SMS + 80-word email version.
The mistake everyone makes
Copying prompts is the start, not the system. If you're pasting the same store details into every chat, you're doing the machine's job for it. The next step is moving your voice, facts, and rules into markdown knowledge files, then packaging repeatable jobs into Skills — so anyone in the store gets the same quality output without knowing the "magic words."
At the workshop, prompt architecture is session one — then we immediately turn your best prompts into knowledge files, skills, and agents. See the full agenda.
Common questions
What makes a good Claude prompt for a dealership?
Four parts every time: role (who the AI is in this task), context (what it needs to know), constraints (rules, tone, what it must never do), and format (exactly what the output looks like). Weak prompts ask. Strong prompts brief.
Can I just copy these prompts?
Yes, that is what the templates are for. Copying is the start, not the system. If you find yourself pasting the same store details into every chat, move your voice, facts, and rules into knowledge files and package repeatable jobs into Skills.
What should a BDC prompt never allow?
Payment quotes, availability promises, and car-sales cliches. Write those as hard constraints in the prompt, and later in your rules file, so every draft inherits them.
How do I get a prompt to sound like my store?
Load three knowledge files (voice, store facts, rules) into a Claude Project and include three of your best real emails or descriptions as examples. Real examples beat adjectives.
From Prompts to Working Systems
Two hands-on days with builders beside you. We are back in Atlanta January 5, 6, and 7, 2027.
Register Now →