AI Won't Save Your Restaurant, But Unified Data Will

5 hard-earned lessons from the operators actually doing the work

There's no shortage of AI hype in restaurants right now. According to Qu's latest restaurant technology report, 73% of operators will be investing in AI by the end of 2026. But here's the uncomfortable truth: only 9% report seeing meaningful results from those investments.

So what separates the brands getting real returns from the ones just burning budget on buzzwords? We sat down with two operators in the trenches—Leon DeVoyan, CTO at Dave's Hot Chicken, and Kevin Flaherty, who leads marketing and culinary at Taco John's—to find out. Their stories cut through the noise in a way the headlines don't.

Here's what they had to say:

1. If your tech stack is fragmented and unstable, your guests feel it first.

Kevin put it plainly: "The area we see it first is usually the guest experience." Different pricing between channels. Promos that work in one place but not another. Inconsistent handoff experiences depending on whether someone ordered through your app, a third-party platform, or walked into the store.

Every one of those breakdowns is a fragmentation tax—and your customer is the one paying it.

For Leon at Dave's, where 40% of transactions are digital, the problem ran deeper. Platforms had dashboards, but none of them talked to each other. Third-party delivery data—a huge chunk of digital sales—was nearly impossible to pull into a single analysis. You can't make smart decisions when 40% of your business is a black box.

The takeaway: Fragmentation isn't an IT problem. It's a guest experience problem and a decision-making problem, and it shows up long before anyone says the word "AI."

2. Loyalty data is lying to you

This was the biggest mic-drop of the conversation. For years, Dave's made business decisions based on loyalty data—because that was the only data they had. Leon's reflection: that approach is "absolutely wrong."

Loyalty guests behave fundamentally differently than non-loyalty guests. They visit more often, spend less per visit, and stick to what they know. Non-loyalty guests visit less, but spend more and are far more likely to try something new. If your entire strategy is built on loyalty data, you're optimizing for one slice of your business and ignoring everyone else.

Kevin echoed the same insight from the marketing side at Taco John's: financial reporting systems were never built for guest behavior reporting, and that distinction is what unlocks better decisions

The takeaway: If you're only measuring the 10–20% of customers in your loyalty program, you're running your business with one eye closed.

3. The cauliflower story: why menu mix lies, too

Leon shared a story that should be required reading for every culinary and marketing leader.

When Dave's launched cauliflower, the menu mix was tiny. By traditional logic, you'd kill it. They did. Sales dipped.

Why? Because cauliflower wasn't just feeding vegetarians—it was anchoring groups. One vegetarian at a table of four was deciding whether all four people walked through the door. Pull cauliflower off the menu, and you lose the whole group.

Even more surprising: first-time customers who came in for cauliflower bought chicken on their next visit. The "low-mix" item was actually a powerful acquisition tool.

Kevin saw the exact same pattern at Taco John's with a seasonal LTO. Small mix, looked like a flop—until the data showed 100% of those users were new guests. "It's no longer an LTO," he said. "It's a new customer acquisition tool."

The takeaway: Menu mix is the most political report in the restaurant business. Without the customer behavior layer underneath it, you'll kill the wrong items and double down on the wrong ones.

4. AI without guest empathy is a one-star review waiting to happen

Leon's voice AI story is the cautionary tale every operator needs to hear. Dave's rolled out voice AI at the drive-through assuming their Gen Z and Gen Alpha customers would love it—after all, they barely talk to each other at dinner. Conversion rates were strong. The technology worked.

They pulled the plug within two weeks. Why? Guests were leaving one-star reviews. They didn't want to talk to a robot.

Same story with vision AI for order accuracy. Worked perfectly in slow periods. Fell apart during the lunch rush, when staff swapped orders around to keep the line moving. Shelved.

Kevin's framing is the right one: AI should free your team to focus on what makes your brand more human, not strip the humanity out of the experience.

The takeaway: Test AI against guest sentiment, not just technical performance. If the technology works but the guest hates it, the technology doesn't work.

5. Clean data first. AI second. Everything else third.

Both Leon and Kevin landed in the same place: there's no shortcut. You cannot put AI on top of fragmented, dirty data and expect anything other than what Kevin called "scaling confusion."

Where they suggest starting:

• Get your POS data right first. It's where the most transactions live.
• Layer in credit card data next. Especially valuable for drive-through brands where the guest is otherwise invisible.
• Bring in loyalty last. Remember it's a slice, not the whole picture.

Once that foundation is in place, AI starts earning its keep in real, measurable ways. Leon's FP&A team now answers complex analytical questions in 10–15 minutes that used to take a week. Kevin's organization is using AI to handle the mundane and repetitive so his team can spend time on the things that actually differentiate the brand.

The takeaway: AI is a force multiplier. If your data foundation is solid, it multiplies the good. If it isn't, it multiplies the chaos.

The bottom line

The brands winning with AI right now aren't the ones with the flashiest demos. They're the ones who did the unsexy work first: unifying their data, killing the silos, and learning to look at their business through the lens of the guest instead of the P&L.

As Leon put it, "Your facts have to be solid before a computer can make decisions for you."

AI isn't the strategy. Clean, connected data is the strategy. AI is what you do once you've earned the right to use it.
‍