A small car dealership can use AI for marketing in four weeks: week one rewrites stock listings and builds a photo routine, week two schedules social posts, week three emails past buyers, and week four requests reviews and measures. Budget four to six hours a week and about $13 to $50 a month.
AI drafts; the lot owns the facts. A listing that overstates condition, mileage or history is a legal and reputational problem, so every week below has a fact check built in. One illustrative lot is followed all the way through.
The lot, and what the plan is for
An illustrative independent lot: 38 used cars on the forecourt, two salespeople, one part-time admin, and an email list of 620 past buyers, about 400 of whom ticked marketing consent. Sales run at 22 cars a month. The owner wants more enquiries from the same stock, not more stock.
The plan is deliberately small. Four jobs, one a week, each ending with something you can count: listings rewritten, posts scheduled, emails sent, reviews requested. After week four you will know which job brought in enquiries, and can then decide what to automate further. If you want the wider picture, how AI can help a small lot over 30 days covers adjacent work such as lead replies and service reminders.
Week one: honest listings and a photo routine
Listings are the highest-value marketing on a lot, since buyers search them before they look at anything else. A good listing is accurate, complete and distinct. Most are thin and identical. AI is good at turning a facts sheet into readable text; it is poor at knowing what is true of a particular car.
Build a facts sheet for every car
For each vehicle, a row with: make, model, year, trim, mileage, fuel, gearbox, colour, number of owners, service history (full, partial, none), known faults or repairs, warranty offered, price, and the vehicle history check result. If you cannot fill a field, leave it as "unknown", and the AI must not fill it either.
Write a used-car listing description of 90 to 120 words from the facts below. British English, plain, specific.
Use only these facts. Do not describe condition, history, accident status, previous use, performance,
fuel economy, safety ratings or finance. If a fact is "unknown", leave it out. No exclamation marks.
End with: "Viewing by appointment. Call or message to arrange."
FACTS:
{facts_row}Illustrative output for a 2018 family hatchback with full service history and 61,400 miles:
A 2018 five-door hatchback in silver, with a petrol engine, manual gearbox and 61,400 miles on the clock. Two previous owners and a full service history. Spacious boot, comfortable for family use, and economical to run. Viewing by appointment. Call or message to arrange.
What to fix: "spacious boot" and "economical to run" are claims the facts did not contain, and "comfortable for family use" is the AI's flourish. Remove them, or replace them with a fact you hold, such as the boot capacity from the manufacturer's sheet. This habit is the same one covered in honest used-car listings and photo clean-up, which goes further on image editing.
The photo routine
Eight photos per car, the same order every time: front three-quarter, side, rear three-quarter, interior front, interior rear, dashboard with mileage, boot, and one close-up of any scuff or mark you want to be open about. Shoot in the same bay at the same time of day. AI photo tools can tidy backgrounds, but never use them to remove dents, scratches or damage; an edited photo that hides a fault is a false description. A fixed routine beats a clever tool.
The week-one finish line
Rewrite the listings for your 10 slowest-moving cars first. Compare enquiries per car over the next three weeks against those you left alone. That small test tells you whether the new copy is worth rolling out.
Week two: social posts that show the stock
Social posting for a lot is mostly a product catalogue with a human voice. Three posts a week is enough: one new arrival, one "just sold" or handover, one useful tip or answer.
- New arrival. Photos 1, 4 and 7 from the routine, the facts line, the price and how to enquire. The AI turns the facts row into a short caption.
- Handover. Only with the buyer's written permission to show them or their car. A buyer shown without consent is a complaint waiting to happen.
- Useful answer. One real question customers ask, such as how a trade-in is valued or what a pre-sale inspection covers.
Illustrative prompt and result for the "useful answer" post:
Write a 70-word Instagram caption answering: "What happens to my car when I trade it in?"
Use only these steps: we inspect it (about 20 minutes), we check its history report, we give a written offer valid for 7 days,
the offer is yours whether or not you buy from us. British English. No emojis. No promises about price.Thinking of trading in? Here is how it works with us. We inspect the car, which takes around 20 minutes, and check its history report. You then get a written offer that stays valid for seven days, and it is yours whether or not you buy from us. Message us to book a time.
That post is accurate because the prompt held the facts. If you let the AI describe how trade-ins "typically" work, it will add steps your lot does not follow.
Schedule the week's three posts in one 30-minute session with whatever scheduler you have. Instagram caps hashtags at five per post, so use two or three that describe the car or the lot. Before anything goes live, run the post through the twelve checks in the pre-post checklist for local business AI marketing, since price and condition claims are exactly what it catches.
Week three: email to past buyers
Past buyers are your cheapest audience. They know you, many will change cars within a few years, and they have friends who are buying. Treat the list with respect: consent first, then relevance.
Segment before you write
- Consented buyers from 2 to 4 years ago. Likely to be considering a change. Message: "your car's value" and a no-obligation valuation offer.
- Consented buyers from the last 12 months. Message: a thank-you, a service tip, and a request to pass your name on.
- Everyone else. No marketing email. If service messages are allowed under the rules where you operate, keep them strictly about the service. Ask an adviser what applies.
Illustrative email to segment 1, drafted by AI from three facts (valuation takes 20 minutes, written offer valid 7 days, no obligation), then edited:
Hi [first name], you bought your [make and model] from us a few years ago. If you are wondering what it is worth now, we are offering free valuations this month. It takes about 20 minutes, you get a written offer valid for seven days, and there is no obligation to buy anything. Reply to this email or call to book a time.
Edits: delete "this month" unless the offer genuinely ends; confirm "free" is true for everyone; add your business address and an unsubscribe link in the footer. Check the footer basics with whoever advises you on marketing rules.
Cost of the email tool
Mailchimp lists Essentials at $13 a month and Standard at $20 a month for 500 contacts, with larger lists costing more. With 400 consented contacts, either tier fits; Essentials is enough for a one-off email, and Standard helps if you want a reminder step that only goes to people who did not open the first. Confirm the current price on Mailchimp's page for your list size.
Week four: review requests and the scorecard
Reviews are the quiet part of dealership marketing. A buyer reads three or four before ringing. The way to more of them is asking every buyer, at the right moment, in a way that follows the review platform's rules.
The rules that matter
- No incentives. Do not offer discounts, free servicing or prize draws for reviews.
- No gating. Do not ask for a private rating first and only send happy customers to the public page. Ask everyone the same way.
- No staff quotas for reviews, which Google's review policy has barred since April 2026.
- Never write reviews yourself, with or without AI.
The request itself
Send it two to three days after handover, when the new-car glow is real but the paperwork is done. A short text or email with a direct link to your review page is enough. Illustrative wording:
Hi [first name], thanks again for choosing us. If you have a minute, an honest review of your experience would help other buyers. Here is the link: [review link]. Thank you, [your name].
AI can help draft replies to the reviews that come in; responding to dealership reviews with AI has samples and rules, including what to do with a poor review.
The scorecard
On the last day of week four, fill in one table. The point is to see which week moved something.
| Job | Measure | Illustrative result |
|---|---|---|
| Rewritten listings (10 cars) | Enquiries per car over 3 weeks, against the 10 untouched | 2.1 against 1.4 |
| Social posts (12 posts) | Messages received that mention a post | 9 |
| Past-buyer email (400 sent) | Replies and valuation bookings | 14 replies, 6 valuations |
| Review requests (22 buyers) | New public reviews | 7 |
These figures are invented for illustration. Your own will differ, and a month on a lot of this size is a small sample: treat the result as a pointer, and repeat the best week next month before drawing conclusions. If the email brought six valuations, and two became purchases at an average gross profit of, say, $1,800 each, that single week would have returned $3,600 against a tool cost of $20. Do your own sum with your real gross profit.
The four weeks on one page, with hours
| Week | Main job | Day by day | Hours |
|---|---|---|---|
| 1 | Listings and photos | Mon: build the stock sheet. Tue: write and test the prompt on three cars. Wed and Thu: rewrite 10 listings and check each fact. Fri: reshoot photos for two cars using the routine | 6 |
| 2 | Social posts | Mon: draft three posts. Tue: run the pre-post checks. Wed: schedule. Thu and Fri: reply to messages, note which post prompted them | 4 |
| 3 | Past-buyer email | Mon: clean and segment the list. Tue: draft and edit the email. Wed: test send to staff. Thu: send. Fri: answer replies | 5 |
| 4 | Reviews and scorecard | Mon: write the request wording. Tue: send to the last 30 days of buyers. Wed to Fri: reply to reviews, fill in the scorecard | 4 |
If a week slips, do not compress the next one to catch up. The plan works because each job is small; stack two and the checks are the first thing to go.
Google Business Profile: the free listing people forget
A lot's profile on a map service is often the first thing a local buyer sees. It deserves twenty minutes in week one. Check the opening hours (including holiday hours), the phone number, the website link and the categories. Add a few real photos of the forecourt and the office. Use AI for the short business description only, giving it three facts: how long you have traded, the type of cars you sell, and your warranty. Keep to what you can honour.
Illustrative description drafted from those facts, and the edit:
Family-run independent dealer, trading since 2011, with a changing selection of used hatchbacks, saloons and family cars, each with a 3-month warranty. Pop in or book a viewing.
"Family-run" came from the owner and is fine; "changing selection" is vague but true. Delete anything that reads like a ranking, and do not add finance language. Posts you add to the profile follow the same twelve-check routine as any social post.
A price-drop post, worked through
A price drop is the cheapest marketing a lot has, and a good test of the fact discipline. Say a 2017 estate has sat for 62 days at $9,995 and you cut it to $9,495. The AI should be told exactly that and nothing more.
Write a 50-word social post announcing a price reduction on one car.
Facts only: 2017 estate, 71,200 miles, full service history, was $9,995, now $9,495, viewing by appointment.
Do not use the words "bargain", "steal" or "must go". No claim about market value.Price reduced: our 2017 estate with 71,200 miles and a full service history is now $9,495, down from $9,995. Viewing by appointment. Message us to book a time.
That is accurate, short and useful. The risk is the version the AI writes without the "do not" line: "an unbeatable deal well below market value". You cannot prove "below market value" unless you have checked comparable listings that day, and keeping that evidence is more work than leaving the phrase out.
When a customer replies to a post or email
Marketing creates replies, and replies create the real work. Decide before week two who answers messages and within what time. A message to a car post that sits for a day is a lost enquiry. If you have no one to watch the inbox, a short auto-reply on social messages helps: "Thanks for your message. We reply within two working hours during opening times. For a quick answer, call [number]." Anything beyond that is a different project, covered in the lead-reply tutorial linked at the end. Keep the rule that AI never confirms a price, availability or finance offer in a reply without a person checking it against the stock sheet.
What it costs, and what you do not need
| Item | Monthly cost |
|---|---|
| Chat assistant for drafting (ChatGPT Plus is $20 a month; a free tier may do) | $0 to $20 |
| Email tool at 500 contacts (Mailchimp Essentials $13, Standard $20) | $13 to $20 |
| Social scheduler, if used | Often free tier, or a few dollars |
| Listing sites and review page | Already paid, or free |
| Total | About $13 to $50 |
What you do not need yet: a dealership-specific AI platform, a chatbot on the site, or a voice agent. They solve different problems, the main one being speed of reply to a lead. Add them once the plan above has shown you where enquiries come from. If a chatbot is next, read the rules to set before launch.
Where dealership marketing goes wrong
The sold car still advertised
A listing stays up after the car is sold, and an enquirer drives across town for nothing. Add "remove listing" to the handover checklist and keep one master stock sheet that every post and listing is generated from.
The price that differs between sites
The AI caption says $11,495; the listing site says $11,995 after a reduction. Use one source of truth for price, and regenerate captions from it, never type prices into a prompt from memory.
The "no accidents" line
An AI writing from "clean history report" turns it into "accident-free". Those are different claims. A report that shows no recorded incidents is not a guarantee of no damage. Use the report's own wording, for example "history check completed, no recorded incidents on the date checked", with the date.
The email to someone who bought a car from a different owner
A list built from old systems sometimes merges records, so a past buyer receives a message about a car they no longer own. Spot-check twenty rows before sending to 400.
After four weeks: what to repeat, what to automate
- Keep the facts sheet and listing prompt as standing routines for every new car.
- Keep the three weekly posts only if enquiries mention them; otherwise cut to two.
- Repeat the past-buyer email quarterly to the right segment, with a fresh fact each time.
- Make review requests part of the handover, not a monthly project.
When the routines are steady, the next gain is usually the speed of the first reply to a web lead, covered in answering a web lead in minutes. If you want someone to map the whole lot against what is worth automating, that is what my AI implementation consultation covers.
Dealership marketing questions
How much time does the four-week plan take each week?
About four to six hours a week for a small lot, with most of it in week one when you set up the listing template and the photo routine. Weeks three and four are lighter, because the email and review requests run on lists you already hold.
Can AI write finance or payment claims in listings?
Leave them out unless your compliance process approves the exact wording. Monthly payments, rates and 'guaranteed approval' language are regulated in most markets. Let AI describe the vehicle; let your finance team or adviser own anything about credit.
Is it fine to email past buyers?
Only those who agreed to marketing, with a working unsubscribe link and your business details in the footer. For others, a service-related message may follow different rules. Ask a qualified adviser what applies, and keep the consent record with each customer.
Should the dealership pay for an AI marketing platform?
Not at first. A chat assistant, a basic email tool and your existing listing sites cover this plan for roughly $13 to $50 a month. Look at dealership-specific platforms once you know which weeks of this plan actually bring in enquiries.
Further reads
- AI for Used Car Sales: Price Each Car Against Local Listings — Price each car against local listings before you post it.
- ChatGPT for Car Dealerships: What Customer Data Never to Paste In — Which customer data must stay out of the chat assistant.
- AI for Auto Dealerships: Where a Small Independent Lot Should Start — Where a small independent lot should start with AI.
- Best AI Platform for Used Car Dealerships: How to Choose — How to judge a dealer platform once you outgrow this plan.
- Mia AI for Car Dealerships: Questions to Ask Before You Sign — A buyer's checklist for a small dealership weighing Mia: handoff, CRM link, after-hours cover, test calls and how to exit the contract.
- AI Tools for Car Sales: Test-Drive Follow-Ups, Quotes and Call Notes — A salesperson's routine for AI follow-ups after a test drive and tidy call notes, with prompts, samples and the lines AI must never write.
- AI-Powered Marketing Tools for a Motorcycle Dealer: Spring Campaigns — A ten-week spring plan for a motorcycle dealer: which AI tools to use for used-bike posts, service reminders and demo days, with list prices.
- ChatGPT Prompts for Car Dealers: 10 for Follow-Ups and Trade-Ins — Ten short ChatGPT prompts for follow-ups, trade-in replies and reviews, each with sample output and a check to run before sending.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: Mailchimp pricing page, Google's review and business-profile policies, platform advertising guidance; confirm finance and advertising rules with your adviser.