← Restaurant Tech Radar
Issue 001 · July 16, 2026 · Free edition

Buy back the hour.

Restaurant tech has to earn its place on the floor. New is cheap. The useful question is whether a tool gives operators more control without making the shift worse.

A restaurant does not need another dashboard just because someone found a new model, payment flow, reservation widget, or ordering screen.

A restaurant needs the hour back.

The hour lost confirming reservations that should have confirmed themselves. The hour a manager spends reconciling delivery tablets instead of watching the room. The prep hour that misses because sales, weather, events, and labor notes never became one useful signal. The server hour lost because the POS, loyalty system, and guest notes tell partial truths.

Restaurant Tech Radar lives between the crew, the guest, and the numbers.

That standard comes from restaurant thinking, not SaaS thinking: respect the physical work; treat hospitality as trust and attention; look at reservations as inventory, waste, pacing, and cash flow; and ask where the money and manager time actually move.

The radar this week

1. Reservations are no longer just reservations

A reservation system used to answer one question: who is coming tonight? Now it can expose the cost of no-shows, shape pacing, collect deposits, and carry useful guest context into service. That only matters if the policy fits the restaurant; a tasting menu, a neighborhood bistro, and a walk-in-heavy bar should not use the same controls.

The owner decision is less about software and more about inventory. Which seats are hard to refill, what does an empty cover cost in contribution margin, and when does a deposit protect the kitchen without punishing a good guest?

Industry case

In Andrew Zimmern's interview with Tock cofounder Nick Kokonas, Zimmern reports that Alinea's profits rose 38 percent after moving to prepaid tickets. That is one high-end operator example, not a benchmark for every restaurant, but it shows why reservation policy belongs in the P&L conversation. Read the interview.

Owner test for the week: Pull last month's no-shows, late cancellations, and same-day party-size changes. Multiply the lost covers by average contribution margin, then separate peak periods from easy-to-refill periods. If the exposure is small, keep the system simple; if it is real, test one policy change before buying more software.

2. Guest memory is becoming a front-of-house system

Hospitality has always depended on memory: regulars, allergies, favorite tables, birthdays, and the guest who wants the check fast. The problem is that memory leaves when someone is off, quits, or gets buried during a rush. Useful guest data has to be short, current, and visible at the moment of service, not trapped in a CRM profile nobody opens.

Two lines can be enough: "Prefers bar seating and usually orders Burgundy" or "Long wait last visit; manager recovered with dessert." Sensitive details need clear rules, and allergy notes still require human confirmation rather than blind trust in a profile.

Industry case · Vendor-reported

A SevenRooms case study says Mina Group used direct reservations and shared guest profiles across its venues, building more than one million profiles and attributing $428,000 in 2021 incremental revenue to automated marketing emails. The numbers come from a technology-provider case study, so treat them as directional; the useful operating pattern is one guest record that can improve service and follow-up across locations. Review the case summary.

Owner test for the week: Pick ten regulars and ask whether a new server could find one useful preference, one service-recovery note, and any critical allergy warning in under 30 seconds. Then check who can edit those notes, when they expire, and whether the restaurant can export them. If the data cannot survive a staffing change, it is storage rather than hospitality memory.

3. AI is useful when it removes manager drag

The first useful restaurant AI jobs are not glamorous. They clean up shift notes, group reviews by recurring complaint, surface menu items where volume rises while margin falls, and answer questions buried in reports. The tool earns a place only when a manager can check the source, correct the answer, and act without creating another nightly reconciliation job.

The industry is moving toward AI inside the system of record rather than a separate chatbot with no restaurant context. That can shorten the path from question to action, but it also raises the stakes when permissions, menu edits, labor data, or recommendations are wrong.

Industry case · Executive-reported

Chipotle rolled out a conversational hiring assistant across more than 3,500 restaurants to answer candidate questions, collect basic information, and schedule interviews. CNBC reported in July 2025 that Chipotle's HR chief said candidates were reaching "ready to hire" in 3.5 days versus as long as 12 days before; the system schedules interviews but does not screen résumés or make hiring decisions. The result is company-reported rather than an independent study, but the boundary is useful: automate manager coordination, not employment judgment. Read the reported follow-up.

Owner test for the week: Review one week of manager notes, reviews, and sales comments and write down what the team learned too late. Choose one repeated question, time the current manual answer, and define the evidence a manager would need before trusting an automated answer. Do not automate the action until the answer survives a week of side-by-side checking.

4. Kitchen signals beat kitchen guesses

Prep is where vague technology dies. A useful system connects item mix, reservations, daypart, weather, local events, waste, 86s, labor limits, and supplier problems to one question: how much should we make? The chef still owns the call; the system has to show enough of its work to earn trust.

Forecasting without waste measurement can make the wrong number look precise. Measure what was prepared, sold, wasted, and unavailable first, then see whether the forecast improves those outcomes over the kitchen's existing method.

Industry case · Peer-reviewed

A 2025 five-site European hospitality study tested AI and computer-vision waste trackers in a resort restaurant, a business caterer, and three hotels. Four sites reduced food waste 23–51 percent while one hotel recorded a 13 percent increase; waste cost per meal fell by as much as 39 percent. The mixed result matters: measurement creates a signal, but staff follow-through and operating context determine whether the signal changes waste. Review the study abstract.

Owner test for the week: Choose five high-variance items and record planned prep, actual sales, waste, and 86s for two weeks. Add one reason code when the plan misses, such as weather, event traffic, reservation change, supplier issue, or cook error. That baseline tells you whether you need better forecasting, better execution, or both.

5. Payments, delivery, and loyalty need a margin audit

Delivery commissions, card fees, loyalty discounts, chargebacks, refunds, promoted listings, packaging, menu markups, and monthly software charges rarely appear in one place. Revenue can rise while contribution margin falls, especially when a higher-volume channel carries a heavier fee and discount stack. Owners need a channel P&L, not separate dashboards that each report a win.

Loyalty deserves the same treatment. Track whether a reward changed frequency or spend after the cost of the reward, rather than counting every returning guest as incremental.

Industry case · Public filing

Domino's reported that more than 85 percent of U.S. retail sales came through digital channels in 2025. In the same annual filing, the company warns that orders placed through aggregators may not have the same store-level profitability as orders placed through owned channels. That is the distinction operators need: "digital" describes how an order arrived, not whether the order was profitable. Read the annual filing.

Owner test for the week: Build one page with gross sales, discounts, refunds, commissions, card fees, packaging, delivery labor, ad spend, chargebacks, loyalty cost, and net contribution by channel. Compare dine-in, pickup, first-party ordering, and each marketplace separately. Give one person ownership of the page and require a monthly explanation for any channel where sales and contribution margin move in opposite directions.

The operator audit: buy back one hour

This week, do not shop for software. For one normal service period, track every avoidable interruption that pulls a manager, chef, host, bartender, or server away from the work that matters.

InterruptionWho got pulled?What caused it?Could tech remove it—or did tech cause it?
Guest called to change reservationHostNo self-serve modification or unclear policyMaybe remove
Expo had wrong modifierServer + kitchenPOS/menu routing issueTech caused it
Manager answered the same private-event question againManagerNo reusable response or intakeRemove
Prep missed because event traffic was invisibleChefForecast blind spotMaybe remove

Circle only interruptions that cost real attention. That is the first buying map—not categories, vendor pages, or whatever got funded this week. Lost attention.

Operator Pro founding beta

Operator Pro does not repeat the free issue with more words. It turns each owner test into a decision model: no-show exposure and deposit policy, guest-data governance and portability, AI answer validation, prep forecast error, and contribution margin by ordering channel.

The Issue 001 companion also includes the 35-point "does this tool pay rent?" scorecard, buying questions, implementation failure modes, and the industry pattern behind each category. No charge today. The beta opens to a small cohort only after fulfillment, support, and cancellation paths are verified.

Request a founding-beta invitation

The bar

Restaurant technology has a bad habit of talking over the restaurant. Good tools do the opposite. They make the room easier to read, make the prep call less of a guess, protect the host stand, help the server remember, and let the manager spend less time reconciling platforms.

If a tool cannot buy back attention, protect margin, or improve hospitality, it does not belong in the stack yet.

Join the free Thursday briefing

Source notes

These sources calibrate RTR's operator standard and support the examples above. Product pricing and capabilities change, and vendor case studies are labeled because they are not independent benchmarks.

Corrections or questions: hello@restauranttechradar.com · Editorial policy