Yum Brands credits voice AI at more than 890 Taco Bell drive-thrus with lower crew turnover

Start with the size of the pool the retention claim has to come out of. If August's 3.5% monthly quit rate held for twelve months, food service would lose about 42 workers per 100 jobs a year ^[14], against about 23 at the 1.9% rate for all nonfarm jobs ^[15], a gap of roughly 19 per 100 ^[16]. (One month annualised is a crude figure, and we use it only to size the gap.) PYMNTS wrote that a drive-thru that holds on to its crew is spared the cost of hiring and training each replacement, a saving that order times and ticket size do not capture ^[22].

Against that pool, the evidence is thin, and the wording gets firmer as it moves up the chain. Omilia's case study reports early indicators ^[18]. Its July 7 release says locations using voice AI had higher retention than those without ^[20]. Yum's Burkhardt went further: "AI is actually increasing retention among our team members in the restaurant because it's making their jobs easier," she said at the Fortune summit ^[6]. None of the three, Yum, Taco Bell or Omilia, has published turnover figures, and every claim so far comes from the companies themselves ^[19].

We think the direction is plausible, since a crew spared the speaker and the complaints has a reason to stay. The counter-thesis is that the gap is real and still too small to cover the software's cost. A store-level turnover figure, set beside a cost per replacement, would settle it, and matched stores quitting near the 3.5% industry rate would prove the claim wrong.

Taco Bell has moved ahead of the numbers. Dane Mathews, its global chief digital and technology officer, called the technology "really, really early" in August 2025 ^[2]. Instead of pulling back, the chain kept expanding ^[3], and the two store counts it has given sit roughly 390 drive-thrus apart ^[13].

McDonald's makes the other case. Fifty labor hours a week per restaurant is 2,600 hours a year ^[17], and Archy is being sold on those hours; the company ended an IBM voice test in 2024 after running it in more than 100 restaurants ^[11]. If a chain turns freed hours into smaller crews, the hours story and the retention story pull in different directions. Omilia's case study describes the freed crew as working the kitchen and the window ^[21].