More than half of your website traffic arrives after you lock up. A shopper with a question at 10 PM has two places to ask: your website or an AI assistant like ChatGPT. Either way, the answer comes from your listing.
That idea came up in the Dealer-Grade AI panel at Golf Carting Expo 2026. Dealer Spike’s Chief Customer Officer, Tamara Burrell, joined the Blackpurl-hosted session to talk about what dealers deal with every day. If you were in the room, this is your recap and your to-do list.
1. Your customer’s AI is already shopping your lot
More buyers start in ChatGPT, Gemini, or Claude before they ever reach a dealer’s website. They aren’t asking for spec sheets. They’re asking whether a cart will get six people to the marina and back without a charge.
An assistant can only answer that from what you’ve published. If your listing says “Onward 4, lithium” and nothing else, the assistant can guess, or it can recommend a dealer whose listing doesn’t make it guess.
The most important AI project this year isn’t a tool. It’s a finished listing.”
Tamara Burrell, Chief Customer Officer, Dealer Spike
Try this prompt. Open ChatGPT, Gemini, or Claude with web search turned on, and ask the question your buyers ask:
I’m looking for a 6-seat electric golf cart with a lithium battery near [your city]. Which dealers have one in stock, and what will it cost out the door?
Run it two or three times, because answers vary. Write down whether you’re named, which dealers are, and what the assistant says about each one’s carts. If a competitor shows up with specs and a price and you don’t, start with your listings. Assistants also read your Google Business Profile and reviews, so check those while you’re there.
2. After hours is where buying decisions start
When 54.1% of shoppers browse outside business hours, after-hours traffic isn’t overflow. It’s where the decision to visit gets made.
The panel set a clear standard for an inquiry that arrives at 9 PM on a Saturday:
- Reply within 15 minutes when you can and confirm that you received it.
- Capture the shopper’s name, phone number, and the cart they asked about.
- Confirm that the cart is still available.
- Offer two times to come see it.
- Say when a person will follow up, and make sure someone does, first thing in the morning.
If a tool books appointments for you, ask one question before you sign: does it write the appointment into your DMS, or into its own calendar? An appointment your team can’t see is one nobody is ready for.
Do this week: Mystery-shop your own site after hours. Around 10 PM, from a phone number and email your team won’t recognize, submit this inquiry on a real cart in your inventory:
Hi, is the [model] on your site still available? I need it to sit six and get to [a spot about 8 miles away] and back on one charge. What’s the out-the-door price, and can I see it this weekend?
That one message tests four things at once: availability, a range question only the listing can answer, price, and scheduling. Score what happens over the next 12 hours:
| Check | A pass looks like |
|---|---|
| Speed | A reply within 15 minutes |
| Specific | It names the cart you asked about, not a generic “thanks for reaching out” |
| Answer | It confirms availability, or offers a close match if that cart is sold |
| Next step | It offers two times to see the cart |
| Handoff | It says when a person will follow up |
| Morning | A person calls or texts before your first appointment, with your question in hand |
| Record | The lead is in your CRM with name, phone, cart, and the full message |
If the automated reply passes and the last three rows miss, that gap is a handoff problem, not a tool problem, so fix who owns the morning follow-up before you shop for new software. Run the test again next month from a different device and a different cart.
3. The top dealers aren’t outspending you. They’re outlisting you.
Among the dealers with the fastest inventory turn, 83.30% of units show a price, units average 10.2 images each, and only 20.92% of inventory is older than 90 days.
Source: The State of the Dealer 2026.
Those dealers sell other vehicles, but golf cart buyers ask the same first questions: what is it, what does it cost, and can I see it? For golf carts, these spec fields carry the most weight:
- Power: voltage (such as 48V), battery type (lithium or lead-acid), amp-hours, and battery warranty
- Range: realistic miles per charge, and what that assumes (passengers, hills)
- Seating: seat count and configuration, such as a rear flip seat
- Build: lift size, plus tire size and wheel size
- Street use: whether it’s a street-legal low-speed vehicle (LSV) with a VIN, seat belts, lights, and mirrors, and its top speed
- Used carts: model year, battery age, and what’s been replaced
Custom and used carts need the most care. A factory model comes with manufacturer specs. A lifted cart with a lithium conversion and a new rear seat only has what you typed in. On a used cart, battery age is often the first thing a buyer asks, because it tells them how soon they’ll need to pay for a new set.
Photos. Shoot the actual cart, not stock images. Get the rear seat, the battery compartment, the dash, and the tires. A short phone walk-around video helps too.
Price. Show the full price and what makes it up: the cart, freight, setup, and doc fee, with accessories listed separately. Check your state’s advertising rules and any OEM minimum advertised price policy. Buyers don’t mind fees nearly as much as they mind surprises.
Try this prompt. Copy the text of one listing into ChatGPT, Gemini, or Claude and ask:
I’m shopping for this golf cart. Using only what’s written in this listing, list the questions I’d still need answered before I’d call or visit. Don’t fill in missing details.
Every question on that list is one your team answers by phone, or one a shopper answers by moving on to the next dealer. Add the answers to the listing.
Do this week: Pull your five oldest carts and check each one against the spec list above, and run the prompt on the two weakest listings.
4. Ask every vendor: is it reading my unit, or the internet?
AI isn’t the problem. The pitch is. Vendors know dealers feel pressure to catch up, and pressure is easy to sell against.
A good tool answers from the spec data on that specific cart. A weak one answers from general knowledge. It will confidently promise a battery or warranty the cart on your floor doesn’t have. That’s worse than no answer, because your customer believes it and walks in expecting it.
Add these questions to your vendor scorecard:
- Is that feature shipping today, or is it on the roadmap?
- How many dealers did that result come from, and are any my size and segment?
- When my information is thin, does it say, “I don’t know,” or does it fill in the blank?
- Does a chat lead land in my CRM with the shopper’s name, phone, cart, and full conversation attached?
- Can I read what shoppers are asking?
- Does it keep price and financing questions with a person?
Dealer Spike holds itself to the same test. Its on-site AI answers from the unit’s description and specs plus a short list of vetted manufacturer sources. It leaves price and financing to a person, because dealers asked for that.
Do this week: Pick a cart on your site with a detail you know is missing, such as the battery warranty, and ask your website chat:
Does this cart come with a battery warranty, and how long is it?
If it gives you a number that isn’t on the listing, you’ve learned something important about that tool.
5. Use AI to find the questions you answer 20 times a month
The most useful thing AI does for Tamara’s team is find patterns no one has time to find by hand. Her team had an AI copilot read 8,142 support cases and group them by what dealers were asking for. Almost a third were things dealers could handle on their own if the answer were easy to reach. One group was 323 dealers asking for a copy of their invoice. Her team made invoices self-serve, and those requests stopped. The best answer was one dealers could find on their own.
You can run the same exercise on your customer questions. Pull last month’s texts, emails, and front-counter notes, and remove customer names and phone numbers. Then paste them into a business-tier AI tool, one whose terms say your data isn’t used to train its models.
Try this prompt:
Group these customer questions by what the customer wanted to know and rank the groups by how often they came up. For the top five, tell me whether a better listing, a posted answer on our website, or a person would have handled it best.
Expect a lot of what comes back to be questions about range, seating, batteries, and price. That’s a list of what’s missing from your listings.
Do this week: Run the prompt on one month of customer questions and fix the top group first.
6. AI is strong on arithmetic and weak on context
The live walkthrough showed both sides. The AI handled the math well, but it jumped to conclusions too fast.
When AI sees aged inventory, it tends to recommend a price cut. If those carts are parked on the back lot where nobody sees them, though, you have a visibility problem, not a pricing problem. Discounting a golf cart nobody has looked at just loses money faster.
AI also can’t know that tariffs raised your cost in August. Give it your cost data, though, and it can flag the carts whose price never caught up.
Two rules came out of the session:
Make it show its work. Add this line to the end of any analysis prompt, then check the estimates yourself:
Before you recommend anything, list which numbers came directly from my file and which ones you estimated or assumed.
Decide what to automate with three questions. Does the task happen often? Is the answer the same nearly every time? Is a wrong answer cheap to catch? If all three are yes, automate it. Trade values, negotiation, and “which cart is right for my family” stay with your people.
And one rule that doesn’t bend: nothing a customer sees goes out without a person reading it first, and a frustrated customer on the phone always reaches a human.
High volume, low complexity goes to the machine. Everything that sells a cart stays with your people.”
Tamara Burrell, Chief Customer Officer, Dealer Spike
Your Monday-morning action plan
- Ask an AI assistant for a golf cart like the ones on your lot and see whether it names you.
- Send a test inquiry to your own site after hours and time the reply.
- Audit your five oldest carts for price, photos, and golf-specific specs, then run the listing prompt on the weakest two.
- Ask your chat tool about a detail you know is missing.
- Run the customer-questions prompt on last month’s messages.
- Ask your website, DMS, CRM, and texting providers three questions: What AI is on my account? What’s on by default? What costs extra?
You may find a feature you’re already paying for and never turned on.
Frequently asked questions
How do AI assistants like ChatGPT decide which golf cart dealer to recommend? They answer from information they can find online, and for inventory questions, your listings play a big role. A listing with a price, photos, and golf cart specs gives an assistant something specific to repeat. A listing with only a model name leaves it guessing or recommending someone else.
What should a golf cart listing include? The full price with freight, setup, and fees; at least 10 photos of the actual cart; voltage, battery type, and battery warranty; range per charge; seating; lift and tire size; and whether it’s street legal. For used carts, add the model year and battery age.
How fast should a golf cart dealer reply to an after-hours inquiry? Within 15 minutes when you can. Confirm the cart is available, offer two times to see it, and have a person follow up first thing the next morning.
Should golf cart dealers use AI for pricing or negotiation? Use AI for math and pattern-finding, such as spotting aged carts or grouping customer questions. Keep trade values, negotiation, and “which cart is right for me” with your people and have a person review anything a customer sees.
See how your listings measure up
The fastest-turning dealers treat merchandising as a routine, not a project: price, photograph, and refresh, the same way every time. Watch The High Performance Recipe on demand, then download the free merchandising checklist and score your own listings.
About the data
The benchmarks in this recap come from The State of the Dealer: 2026. The report is based on observed behavior from January through December 2025 across more than 6,800 dealers in powersports, marine, trailer, agriculture, heavy truck, heavy equipment, and outdoor power equipment. The vehicles are different, but the lesson carries over to golf carts: the listing is where the sale starts.
About the speaker
Tamara Burrell is Chief Customer Officer at Dealer Spike, where she leads the customer experience teams supporting more than 6,800 dealers. She spends her days on what dealers need and how quickly they get it.
