Dealerships Got Faster in 2026. Buyers Got Less Satisfied. New Data Says Speed Was Never the Bottleneck.

Two numbers that landed on the same day

On August 3, CDK Global published its July read on the car-buying experience. Eighty-one percent of buyers said the process was easy - barely above June's 80%, and well below July 2025's 86% and the 2025 average of 84%. With the single exception of the test drive, every stage of the purchase process was flat or down.

The same day, CBT News reported on separate CDK data pointing at a narrower target: the phone. Forty-seven percent of sales appointments and 61% of service appointments are still booked by phone, 29% of service shoppers said they had difficulty scheduling by phone, and the average dealership service hold time sits at 9.3 minutes.

Neither of those is a story about technology budgets. Read alongside a third number, they become one. CDK's 2026 Friction Points Study reports dealer AI adoption climbing from 28% to 39%, with deal completion times faster than they have ever been.

Faster store. More AI. Less satisfied customer. That combination is worth sitting with before the next vendor demo.

The myth: Satisfaction follows technology adoption - buy the AI, watch the scores climb. The data: In CDK's 2026 Friction Points Study, dealer AI adoption rose from 28% to 39% and deal completion times hit record speed, while Net Promoter Score for buyers at the dealership fell from +48 to +29 and online-only shoppers fell to zero. Two of those three lines moved the way dealers wanted. The one customers actually vote with did not. Source: CDK Global, 2026 Friction Points Study.

Two credible studies, opposite headlines

Here is what makes this genuinely hard to reason about. Two of the most-cited data sets in retail automotive currently say close to opposite things.

A reasonable manager could look at that and conclude the data is noise. It is not. The two studies measured different things, and the difference is the whole point.

Cox measured the clock: time in the store, time researching, hours to complete a deal. On those measures the industry genuinely improved. CDK measured friction - specifically, the moments where a customer wanted something and had to wait, ask twice, or negotiate for it.

CDK's own July detail confirms the split inside a single survey. The time measures were the good news: 55% of buyers said the process took about as long as expected, a record high against a 2025 average of 49%. The negotiation measures were the bad news: satisfaction with negotiating the final price came in at 59%, down four points from May, and agreeing on trade-in value at 50%.

The store got faster. The customer got less happy anyway. Speed was not the binding constraint.

Where dealership AI has actually gone

That reframe matters because it tells you what most 2026 AI spend has and has not touched.

Reynolds and Reynolds surveyed more than 500 dealership employees for its State of AI in Automotive Retail Q1 2026 report and found 57% of dealership personnel using AI in some capacity, rising to 70% among executives and dealer principals. The specific wins respondents named are instructive.

Every one of those is a production task. They are things the dealership makes: copy, reports, replies. AI is very good at them, the time savings are real, and 77% of respondents rated their current AI tools a 5 or better out of 10.

Now compare that list to what CDK's buyers complained about: price transparency, negotiating room, waiting, and not being able to reach someone. Those are access and trust problems. They are not solved by producing the same output faster. A vehicle description written in 40 seconds instead of 12 minutes does not shorten a 9.3-minute hold.

That is the gap. Dealerships bought speed. Customers were asking for access.

The phone is the cleanest unaddressed access point

Of all the friction CDK identifies, the phone carries the least ambiguity - because the customer has already declared intent by dialing, and the failure is binary and measurable.

The numbers are not marginal. Nearly half of sales appointments and a clear majority of service appointments still start with a call, in a year when digital retailing is more capable than it has ever been. And 29% of service shoppers, close to one in three, reported difficulty getting an appointment booked over the phone.

CDK also reports that 81% of dealership sales teams name improving conversion from existing opportunities as their priority. It is difficult to square that priority with a nine-minute hold. The opportunity is already in hand; it is being dropped at the point of contact, not the point of pitch.

We have written before about what missed calls actually cost a store and about service call handling in fixed ops. The 2026 data does not change that arithmetic. It changes the context around it: this is now the friction point that has visibly failed to improve while everything measurable about internal speed improved.

One wrinkle worth noticing

There is a detail in the Cox study that cuts against the simple AI-helps narrative in an interesting way.

Cox found that 19% of all buyers, and 25% of new-car buyers, used AI tools such as ChatGPT or AI search summaries while shopping. Among mostly-digital buyers, those who used AI reported 84% satisfaction with the overall shopping experience against 71% for non-AI users, and 81% against 67% on trusting they got a good deal.

That is a real effect. But note whose AI it was. The satisfaction lift Cox measured came from AI the customer brought to the transaction - tools that made pricing and product information legible before they ever walked in. It is not evidence that the AI a dealership buys raises satisfaction.

Both things can be true at once: customer-side AI is raising transparency expectations, and dealer-side AI has mostly been pointed somewhere else.

Four questions worth answering before the next demo

None of these require a purchase. All four are answerable from a phone system and a CRM this week.

1. What percentage of inbound calls go unanswered, broken out by hour and by department? An aggregate answer rate hides the problem. The leak is almost always concentrated in specific hours - lunch, the last hour of the service drive, Saturday mid-morning.
2. What is your abandon rate at 60 seconds and at three minutes? If the reported average service hold is 9.3 minutes, the relevant question is how many of your callers never make it that far.
3. Of answered calls, what share end with neither a booked appointment nor a scheduled follow-up? This is the handled-but-not-converted bucket, and in most stores nobody owns it.
4. What share of your AI spend touches a moment when a customer is waiting? Add up the line items. If the answer is close to zero while your friction scores fall, the mismatch described above is your mismatch too.

The measured read

The honest summary is that dealership AI in 2026 has been a real operational success and a customer-experience non-event, and those two things are not contradictory. The industry got meaningfully better at producing work. It has not yet gotten meaningfully more available.

Two caveats belong on that read. First, CDK sells communication and CRM products and its friction research is not disinterested; the findings line up with independent buyer surveys, but the framing serves a commercial interest and should be weighted accordingly. The same applies to Reynolds and Reynolds. Second, one month of ease-of-purchase data is one month - CDK itself notes the series is neither linear nor obviously seasonal, with a record high of 92% in January 2025 and a record low of 66% ten months later.

What would change this read: a run of months where satisfaction and access measures rise alongside AI adoption instead of diverging from it. That has not happened yet. Until it does, the more defensible position is that the next increment of AI value at a dealership sits at the contact points customers are already trying to use - and the phone is the one with a number attached to it.

If you are working through which of those contact points your store actually leaks at, and what an AI phone agent would and would not fix, that is the conversation the team at Carbuki has with dealerships.

Sources

• July Car Purchase Process Gets No Summertime Boost - CDK Global, August 3, 2026 (ease-of-purchase, negotiation, trade-in and time figures)
• CDK finds phone friction persists at dealerships as AI adoption grows - CBT News, August 3, 2026
• Missed Dealership Phone Calls Are Costing You Customers - CDK Global, July 28, 2026 (47% and 61% booking-by-phone, 29% scheduling difficulty, 9.3-minute hold, 81% sales-team priority)
• CDK Releases the 2026 Friction Points Study - CDK Global, January 29, 2026 (NPS +48 to +29, AI adoption 28% to 39%)
• Automotive Retail is Reaching a Tipping Point: 57% of Dealership Staff Now Using AI - Digital Dealer, January 26, 2026, reporting the Reynolds and Reynolds State of AI in Automotive Retail Q1 2026 survey
• Efficiency, Digital Tools, and AI Drive Car Buyer Satisfaction to Record Highs, Cox Automotive Study Finds - Digital Dealer, January 19, 2026, reporting the Cox Automotive 2025 Car Buyer Journey Study