The Phones Got Smarter. The Spreadsheets Didn't Notice.

One data point this week does most of the talking, so let's not bury it.

The Big Story

Cox Automotive's new AI in Auto Retail Tracker put a number on something every service counter and sales floor already sensed: dealers bought the AI, but they didn't get the receipts. 82% of stores now have some AI phone tool running — call answering, lead qualification, appointment setting, the usual pitch deck. Only 22% can show a sales number that actually moved because of it.

That's a 60-point gap between "we deployed something" and "we can prove it worked." If you're an operator, that gap should bother you more than the adoption number impresses you. Buying AI phones isn't the achievement. Vendors will happily count you as a win the moment the contract's signed, whether or not a single extra unit rolled off the lot because of it.

The tracker did surface exceptions, and they're worth studying precisely because they're rare: Paragon Honda, Bob King Mazda, and Etheridge Ford all have documented case studies showing where the money actually shows up — not "engagement," not "call volume," but appointments that convert and deals that close. What separates them from the other 78% isn't a better AI vendor. It's that they built a way to measure the thing before they bought the thing, then held the tool to that number every month.

What this means for your store: if you can't currently draw a straight line from your AI phone system to a sales outcome, you don't have an AI strategy — you have an AI subscription. The fix isn't necessarily new software. It's assigning someone to track outcomes weekly against a baseline you set before go-live, the same discipline you'd apply to a new BDC hire or a new pay plan. If nobody's watching the number, the number isn't moving, no matter what the vendor's dashboard claims.

Also This Week

• Voice AI — The Cox Tracker's finding that adoption has outpaced measurement isn't a one-store problem, it's an industry-wide one. Most groups rolling out AI answering services still have no defined "success" metric beyond "the phone got picked up." Picked up by what, doing what, is the question nobody's answering yet.
• CRM — The stores in the case studies all had one unglamorous thing in common: clean CRM attribution tying AI-handled calls to actual appointment records. Without that plumbing, any ROI claim from a vendor is a guess dressed up as a chart.

Worth Watching

Expect more vendors to start citing Paragon, Bob King, and Etheridge by name in their sales decks — that's what happens the moment three credible case studies exist in a sea of unproven ones. The question worth asking back: can you show me my numbers, not theirs. If the answer is a demo instead of a report, you already know where you'll land in next year's tracker.