Reviews are a complaint log, not a scoreboard
Most dealers treat the review page like a report card. Star average goes down, everyone gets a speech about "delivering a five-star experience." Star average goes up, nobody says anything.
That's a waste of the best free data you have. A one-star review is a customer writing you a defect report for free, with a timestamp and a department attached.
The problem is that nobody turns it into a process change. Somebody responds publicly — "We're sorry to hear about your experience, please reach out to our customer care team" — and the complaint dies there. Next month, three more people write the same thing.
Here's the loop that actually fixes it: read reviews in batches, find the repeat, name the step in your sales process that's broken, assign one person to change it, and then go listen to calls to see if the customer experience actually changed.
Step 1: Batch your reviews and code them by process step
Reading reviews one at a time as they come in makes every review feel like an isolated bad day. Read 30 at once and the patterns jump out.
Once a month, pull every review from the last 60 days — Google, DealerRater, Cars.com, Facebook, whatever you're on. Don't just read the one-stars. Three-stars are where the real process problems hide, because a three-star customer isn't angry, they're describing something that annoyed them factually.
Now code each one. Not by sentiment. By which step in the deal it happened at:
- First contact / responsiveness
- Appointment setting and arrival
- Vehicle presentation and demo
- Trade appraisal
- Numbers and negotiation
- F&I
- Delivery
- Post-sale / service handoff
You're not trying to be scientific. You're trying to answer one question: where in the process do people keep getting hurt?
A real example of what this looks like when you code it:
"Drove 45 minutes because they told me the truck was there. It sold the day before. Nobody called me."
That's not a rudeness complaint. That's an appointment confirmation defect.
"Salesperson was great. Then we sat for two hours waiting for finance and nobody told us anything."
That's not an F&I complaint about products. That's a wait-time communication defect, and it belongs to whoever owns the desk-to-F&I handoff.
"They ran my credit without asking."
That's a disclosure language defect at the numbers step, and it's the kind of thing that shows up in five reviews before anybody acts.
This kind of dealership review analysis takes about 45 minutes a month. It's the cheapest diagnostic work you will ever do.
Step 2: Find the repeat, and only work the repeat
You'll finish coding and have eight categories with something in them. Ignore six of them.
Pick the one with the most entries in the last 60 days. That's your fix for the month. One.
The reason to work only the repeat is that a single bad review is usually a bad day — a rep going through something, a customer who was never going to be happy. Four reviews saying the same sentence is a process that permits the failure. Those are different problems, and only the second one is worth changing a process over.
A useful gut check: could a good employee, doing their honest best, still cause this complaint? If yes, it's a process problem. If no, it's a coaching or a people problem, and it goes to that rep's one-on-one instead of your process list.
"Nobody called me back for three days" — a good employee could absolutely cause that if there's no rule about who owns a lead after the first touch. Process.
"He was condescending to my wife and only talked to me" — no. That one's a conversation with the rep, today.
Step 3: Turn the complaint into a sentence someone says
This is the step everyone skips, and it's the reason dealership reputation management usually accomplishes nothing internally.
A fix like "improve communication around wait times" is not a fix. Nobody can do it, and you can't tell whether they did.
A fix has to be a specific behavior, at a specific moment, with words attached. Take the two-hour F&I wait:
Bad fix: "Sales team, keep customers updated."
Real fix: When the deal goes to the desk, the salesperson says this before they walk away:
"Here's what happens next. I'm taking your paperwork to our finance office. That's usually 30 to 45 minutes, sometimes longer if we're busy. I'm going to come back and check on you at the 20-minute mark either way, even if I don't have news yet. Can I get you a water while you wait?"
And then the salesperson comes back at 20 minutes. That's it. That's the whole fix.
Now the appointment one:
Bad fix: "Confirm appointments better."
Real fix: Every confirmation call the day before includes a physical vehicle check first, and the rep says:
"I walked out and put my hands on the blue F-150, stock number 4471, this morning — it's here and it's yours until 2 o'clock tomorrow. Are we still good for 2?"
You can hear the difference. The second one is a behavior you can score on a recorded call. The first one is a feeling.
Step 4: One owner, with a name
Write it down like this:
| Complaint pattern | Count (60 days) | Process step | Fix behavior | Owner | Check date |
|---|---|---|---|---|---|
| Car wasn't there | 5 | Appointment confirm | Physical unit check + stock number in confirm call | Marcus (BDC) | Nov 14 |
"The sales team" is not an owner. "Marcus" is an owner. One name per fix, and that person's job is not to hope — it's to teach the words, model them once, and report back on the check date.
If you're the owner of the store and you can't name who owns the fix, you own it. That's fine. Just say so out loud.
Step 5: Verify with calls, not with the review page
Here's where most dealers get burned. They make the change, and then they watch the review page to see if it worked.
That's a terrible verification signal for three reasons. Reviews lag the experience by weeks. Only a small slice of customers write them. And the customers you helped most are often the quietest.
Verify upstream, in the conversation itself.
Two weeks after the fix, pull 10 calls at the relevant step. For the appointment fix, that's 10 confirmation calls. Listen for exactly one thing: did the rep name the stock number and say they physically confirmed the vehicle?
Score it as a raw count. "7 of 10." Not "generally improving."
Then you know what you're dealing with:
- 8–10 of 10 — it stuck. Move to the next complaint pattern.
- 4–7 of 10 — partial adoption. Usually a wording problem. The reps who skipped it will tell you why in about 30 seconds if you ask.
- 0–3 of 10 — nobody was actually taught. This is on the owner of the fix, not the floor.
That call sample is your real customer experience process measurement. The review page is the lagging confirmation that shows up 60 days later.
What the loop looks like over a quarter
Month one: coded 30 reviews, five said the car wasn't there, assigned the confirmation script to the BDC lead, pulled 10 calls two weeks later, got 6 of 10, retaught, got 9 of 10 at week five.
Month two: coded again. "Car wasn't there" went from five mentions to one. New top pattern is trade appraisal — three customers said the number changed after they got there. New fix, new owner.
Month three: same drill.
Three fixes a quarter is twelve a year. Twelve real defects removed from your customer experience process, each one verified by listening to what reps actually said, not by reading what management hoped they'd say.
That's how car dealer reviews stop being a scoreboard you react to and start being a work queue you clear.
If pulling and scoring those verification calls by hand is the part that keeps falling off your plate, that's exactly the kind of thing MoreSignal was built to handle — flagging whether the new language showed up in the calls, so your check date has an actual number attached to it.
Either way, the discipline is the same. Find the repeat. Write the sentence. Name the owner. Go listen.