CSI isn't a survey problem, it's an expectation problem

A low score almost never means your people were rude. It usually means someone earlier in the chain made a promise that someone later in the chain couldn't keep.

The ad said "no hidden fees." The BDC said "your vehicle is here and ready." The rep said "an hour, tops." Delivery took three. Finance added a doc fee nobody mentioned. Each person did roughly their job. The customer experienced one broken promise after another.

So when your dealership CSI score drops, don't start with the score. Start with the chain. Six links, in order: advertising, BDC, CRM handoff, showroom, delivery, post-sale. Your job is to find the link where the customer's expectation first diverged from what your store actually does — and then fix the promise, not the person.

Run the audit as one customer, not six departments

The mistake most stores make is auditing each department separately. Advertising reviews ads. The BDC manager reviews calls. The sales desk reviews deals. Everyone passes their own check, and the customer still leaves unhappy.

A customer expectation audit works the other way. You pick five recent deals — ideally two with low survey scores, two with high, one that never got returned — and you walk every one of them end to end as a single story.

For each deal, collect:

  • The ad or listing the lead came from (screenshot the price and the disclaimer)
  • The first inbound call recording and the text thread
  • The CRM notes at the moment of handoff to the sales floor
  • The showroom log entry and the desk's worksheet
  • The delivery checklist and time stamps
  • Every post-sale touch through day 30

Then read it in one sitting. Out loud, if you can, with the salesperson in the room. You're looking for one thing: the first sentence where what the customer was told stopped matching what the store was going to do.

That sentence is your defect. Everything downstream is a symptom.

Link 1: Advertising sets the price expectation

Most CSI damage that shows up in finance was created in an ad three weeks earlier.

Check the actual listing, not the media plan. Line up the advertised price against the deal jacket:

  • Does the advertised price require financing with a captive lender? Is that stated in a size the customer would read on a phone?
  • Are dealer-installed accessories in the price, or added later?
  • Does the photo set match the actual trim on the lot?
  • Does "$0 down" survive a 620 credit score?

If an ad price requires four stacked rebates, decide right now who says that out loud and when. The answer should be "the BDC, on the first call," not "finance, at hour three."

"That price is with the loyalty and military rebates. Do either of those apply to you? If not, your price is $2,400 higher, and I'd rather you know that now than sitting at my desk."

A rep who says that on the phone loses a few appointments and saves your survey scores.

Link 2: The BDC promises things the floor has to deliver

Pull twenty BDC calls and listen only for promises. Not tone. Not the script. Promises.

Common ones that break:

  • "It's here on the lot" — when it's in transit or at a sister store
  • "Ask for me when you get here" — when the BDC agent isn't on the floor
  • "We'll have your trade number ready" — when nobody appraised it
  • "Should take about 45 minutes" — for a process that runs two hours on a Saturday

Score each call on a simple binary: did the agent promise anything the store cannot reliably deliver? If more than two in twenty do, you don't have an agent problem, you have a script problem.

Then fix the language. "It's here" becomes "I'm looking at it in our inventory — let me physically confirm before you drive over, and I'll text you a photo with today's date."

Link 3: The BDC handoff is where context dies

This is the link almost nobody audits, and it's the most expensive one.

The customer spent nine minutes on the phone explaining they need a third row, they're upside down about four grand, and they can only come in after 6. Then they arrive and the salesperson opens with "So, what brings you in today?"

That single question tells the customer nobody talked to anybody. Everything after it is uphill.

A clean BDC handoff should carry five things into the showroom before the customer walks in:

  1. Stock number or specific vehicle discussed
  2. Trade: year, model, payoff if known, whether a number was quoted
  3. The stated reason for buying, in the customer's words
  4. Any promise made (price, timing, who they'd meet)
  5. Constraints — payment ceiling, time available, who else decides

Audit this by reading CRM notes cold and asking: could a rep who's never seen this deal greet the customer correctly from these notes alone? If the note says "coming in at 6, wants Explorer," the answer is no.

Have the rep open with the handoff instead:

"You talked to Dana about the white Explorer with the third row, and she said you'd be tight on time tonight. I've got it pulled up front and the keys in my pocket. Before we drive it, tell me about the Pilot you're trading."

Link 4 and 5: Showroom and delivery are where time gets stolen

Customers rarely rate you on price. They rate you on time and surprises.

Time-stamp five recent deals from greeting to keys. Then compare that to what the customer was told. If your store's real median is 2 hours 40 minutes and your reps say "about an hour," you have a chronic CSI defect with no bad actor in it.

Two fixes, both boring:

  • Quote the real number, with a checkpoint. "Realistically, two and a half hours today. I'll tell you where we are at every step, and if we're running long I'll say so."
  • Set a mid-process update rule. Every 20 minutes in finance wait, someone physically goes to the customer and says something true.

At delivery, audit whether these actually happened, not whether the box got checked:

  • Phone paired, and the customer did it, not the rep
  • Second key handed over or a written date for when it arrives
  • We-owe items in writing, with a name and a day
  • Service intro — an actual human, not a pointed finger
  • First payment date said out loud

Ask the customer one question at the curb: "Is there anything today that went differently than you expected?" Write the answer in the CRM verbatim. That's your CSI early-warning system, and it costs nothing.

Link 6: Post-sale follow-up either confirms or contradicts

The survey usually lands before your first follow-up call. That's backwards.

If a we-owe is outstanding on day three and nobody has called, the customer answers the survey about the we-owe. Your follow-up cadence should beat the survey, not trail it.

Check: on your last ten deliveries, what was the first outbound touch, and did it happen before the survey went out? If your CSI dips land on deals with open we-owes, you've found your root cause and it isn't the salesperson's attitude.

Turn the audit into two assignments, not a meeting

Five deals will give you more findings than you can fix. Pick two.

Write them as a defect, an owner, and a date:

"BDC is saying 'it's on the lot' without confirming. Dana owns rewriting the confirmation line by Friday; I'll re-score twenty calls the following Monday."

"Reps quote one hour on Saturdays. Real median is 2:40. New standard: quote 2:30 with a 20-minute update rule. Floor managers own the updates; I'll time-stamp five deals next weekend."

Then re-run the same six-link walk in 30 days on five new deals. If the same link breaks twice, you're looking at a standard nobody agreed to, not a person who forgot.

Dealership customer satisfaction is mostly the absence of surprises. Every surprise was a promise made by someone who wasn't there when it came due. Listening across the whole chain — the calls, the texts, the handoff notes — is how you find who made it. Tools like MoreSignal score those conversations against a rubric so you can see the broken promise on the recording instead of guessing at it from a survey comment. But the audit works with a notepad and a Saturday, too.