Technology vs. Human Judgment

Anthony Valletta: A few episodes ago, I told you people do not want more screens between them and a good night. A bunch of you heard that as me being anti-technology. So I want to be very clear about who is talking. We are building and piloting AI right now on so many different facets, including building out our own internal AI data warehouse for GMs to have instant information at their fingertips, no more cross-referencing multiple reports. We're spending time on video AI within our security camera technology to understand guest behaviors, service standards, and much more. There's happening more vendor demos this year than I want to admit out loud. I'm not the guy who wants to go back to paper tickets. I'm probably one of the most aggressive tech buyers you will hear from this year. And I am telling you that most of the technology being sold into restaurants right now is not built to make your guests' night better. It is built so you can stop thinking about your guest night. That is the actual product, not efficiency, permission. And it works. It works so well that your PL will look great for about four quarters while your guest quietly decides you're not worth the drive. I am Anthony. This is the return. There are two kinds of restaurant technology, and this industry keeps talking about them like they are the same thing. The first kind takes work off your team. Invoices, inventory counts, scheduling, admin, reporting, compliance, all the stuff that drags a manager into the office at seven o'clock on a Friday night. When they should be on the floor. Automate every bit of that. Go as hard as you can. I'll even help you build it. The second kind takes the thinking off your team. Forecasting, guest insight, marketing decisions, what to promote, who your best guess is, where your business is coming from next week. The second kind is not removing labor, is removing judgment. And the pitch sounds identical to the first one, which is exactly why so many operators buy the second thing. While telling themselves they bought the first. So here is the whole episode in one question. What will my people stop doing and what will they do instead? If you cannot answer the second half of that out loud, you did not buy a tool. You bought an exit from the part of this job that actually matters. Three things, same as every week the disagreement, the cut, and the lie. Here is the position a smart person will fight me on. We've tested seven different forecasting tools. Seven. And we're in talks of one more right now. Some of them are good, a couple were better than good. They were genuinely impressive. Not one of them could beat our team. I'm not going to insult you with our people are special. That's the laziest sentence an operator can say. Our team wins for a specific reason. They know things the model cannot see. They know the main road is closing starting Thursday. They know the high school moved graduation. They know the hotel down the street is between flags and running half empty. They know the two servers who make a Tuesday feel like a Friday are both off next week. They know the place two doors down just reopened after a remodel and is buying the neighborhood back with cheap drinks. None of that is in the data. All of it moves the number. Now here's the part that should actually scare you, and it is not the tool that gets it wrong. A good one is right most of the time. That is the problem. Being right most of the time is how it earns your trust. And once it has your trust, you stop doing the thing that made you good in the first place. You know that feeling where you walk in the building and you can feel Saturday before it happens? You look at the book or the patio or the weather, who's on the floor, and a number shows up in your head before you open a single report. That is not instinct you were born with. That is thousands of reps of guessing and then finding out you were wrong. It is trained. And anything trained goes soft when you stop using it. Six months into a good forecasting tool, your managers do not guess anymore. They receive. They are not wrong more often, they are just not practicing. And a manager who stopped practicing is a manager who stopped noticing, because noticing was the whole input to the guess. Then you get the day the model has never seen. Weather turns, a concert reschedules, something goes viral, a convention moves across town. You know exactly what the day looks like in your marketing. The tool misses. Fine, everybody misses. But nobody in your building catches it because catching it was a skill that atrophied on your watch. So now you are many covers deep on a night you scheduled light, the kitchen is buried, your host is apologizing to a family with two kids past bedtime, and a guest who waited 50 minutes for an entree is deciding whether you get another shot. That guest does not know a piece of software failed. They just know you are not ready. Which is the same conclusion they would have drawn 100 years ago. The counter is real, so I'll say it for you. Most operators are not good at forecasting. Most operators are drowning, they are guessing badly. And the tool that is right 90% of the time may be a massive upgrade over a manager who is right 70% of the time and has no time to think about it. That is true. I believe that. If you are that operator, buy the tool. But if you're trying to be the best in your market, understand the trade you are making. You are outsourcing the exact muscle that separates you from the guy across the street who is also about to buy the same tool from the same rep. And if you do buy it, at least run it in the right order. Make your manager commit their number first. Write it down, then open the tool, then talk about the gap. Now it is a sparring partner instead of a replacement. And that muscle keeps working. Costs you 90 seconds a week. Almost nobody does it. The cut is the same one the whole industry made and is the biggest one we have made in my career. Labor got expensive, commodities went up, rent went up, the PL stopped working. So the industry did the rational thing and looked for a way to serve the same number of people with fewer people. Kiosks, tablets, app ordering, order on a screen, carry your own tray, bust your own table, and by the way, screen still asks you for a tip. The math was airtight. Order accuracy approved, average check went up because the screen upsells without ever getting embarrassed. Throughput held, and in some cases improved. Every number in the deck moved the right direction. And here's what happened to me as the guest. And it happened to you. I paid $25 at a known QSR for burger, fries, and a drink. I did not speak to one human being in the building. Nobody. Price up, contact down. Sit with that trade for a second. I am not just paying more. I am paying more, doing more of the work myself, and getting less reason to be there instead of anywhere else. That's not a strategy, that is a countdown. By the way, this is how Chile's really started winning. Their value proposition was cheaper than QSR. Full service, same, if not better quality. That's a compelling story. Now here is the part that should get your attention. Because the smartest brands in the world are quietly walking it back. Some the biggest names in FastCasual are adding guest facing labor back into their buildings on purpose. The exact thing the technology was brought to eliminate. That is not the technology failing. The technology did precisely what it promised. What failed was the assumption underneath it, which was that hospitality is a cost. It never was. Hospitality is the reason you get to charge what you charge. And I want to be fair, because there's an honest version of this cut. If you take the price down along with the contact, the trade is clean. Nobody expects a moment from a vending machine, and nobody pays vending machine prices for a night out. The break happens when you keep charging restaurant prices for a self-service experience. Then you have moved yourself into a fight you cannot win against every place that still has a person at the door. So let me put our own tab on the table because I'm not going to sit here and tell you what to do without it. We pulled our QR code ordering from the front stage of our ordering and experience. We'd have started at the start of 2025 that we had to lean more into the experience. It was going to cost us a ton of cash in the tech stack transition, easily 100 to 150 basis points on our PL, and a whole ton in training our team to a new style of service. The QR was efficient, helped our bottom line. Some guests liked it. Some still do, but most didn't. We were losing 10% year over year over year of guests who hated the QR and never came back. And that offset of those who enjoy it wasn't enough to combat those losses. Our guests told us in a hundred small ways that it made us feel like less of a place worth going to. We built it for the right reason and at the right time, and we kept it for too long after. We had a portion size to more than half our tacos and didn't take price to cover it. Going on almost two years now, we have not taken price. Intentionally, and a stretch where the industry has continued to take price on top of major price increases over the last three to four years already. We put labor back in on purpose and give up a full point of profitability for the first half of 2026 and most of 2025, so there would be more people in the building actually taking care of people. And every piece of technology we've added in the same window sits behind the guest. Better prep forecasting, third-party alerts, and AI reporting, amongst many more. Not one guest facing automation. It is not an accident. That is a rule we wrote down. Are we there? No. Ask my operators. They'll hand you a list. But those are the right investments, and I would make every one of them again. Everyone says it. It is in every deck, every press release, every vendor second slide, and it is a lie. Not because it could not be true, but because almost nobody does the second half. You free up four manager hours a week. Where do those four hours go? Be honest. They go to more email. They go to another report somebody built because the data finally existed. They go to covering a shift you didn't feel. Or they go to the schedule as a labor cut. Because the whole business case you sign to the tool pays for itself an hour saved. That is the tell. If the savings were promised to the PL, the guest was never going to see them. The freed time was spoken for before you turned the thing on. And even when we do try to point technology at the guest, watch what actually happens on the floor. Personalization is the one I'm most excited about on paper. Your guest walks in, your team already knows they come in every other Thursday, they like the salad chopped, they had a bad night in March that we made right, their anniversary is next week, 15 years ago. That lived in one bartender's head. And it walked out the door when he quit. Open table cornered this market for decades with guest and reservation notes. Now I can live in a fully automated AI system. That is real progress, and I want it. Then a 19 year old server, four weeks in, opens the table on a handheld and sees five facts about a stranger. And nobody has ever taught them what to do with that fact. So they do what any of us would do. They read it out loud. Hey, I saw you like your salad chopped, so I had them chop it. Which looks fine typed into a training deck and lands like a mild privacy violation when a human says it to your face. Well say nothing at all. The moment's wasted. Or worse, they perform it. The guest hears the performance, and now your guest is not charmed. They're being marketed to during dinner. The tool delivered the information perfectly. The moment still died. Because information was never the hard part. The hard part is delivery, timing, reading whether the table wants to be known tonight or wants to be left completely alone. Some guests light up when you remember them. Some guests are on a first date and prefer you never seen them before in your life. Telling those two apart in three seconds is a craft, and no vendor on earth is selling it because it cannot be shipped. It has to be taught in person by someone who is good at it, over and over. Genuine is the whole word in that sentence. It is the only word that matters. A guest can tell the difference between somebody who noticed them and somebody who read about them every single time. You felt it. Somebody use your name like they meant it. And somewhere, somebody else used your name because a screen told them to. Same data, opposite outcome. The only variable is a person and whether you bother to prepare them. And the last defense I get on this one is that AI will eventually coach the 19-year-old too. Maybe, probably. Honestly, for the informational part, it can already tell them what to say. What it cannot do is stand at the table with enough time to actually see the person sitting there. And if you cut the labor that puts a human in that spot, the coaching does not matter because there's nobody left to coach. The constraint was never information. It was always whether a prepared human was standing there and not underwater. My cell number is on the receipts. Same ask as last time, and this episode is the one it was built for. Send me the cost cut you've made that you suspected your guests can feel. The one that saved you real money and may have quietly cost you something you cannot see on a report yet. If it is a piece of technology, even better. Tell me what your team stopped doing when you turned it on. And tell me honestly where that time went. Anonymous is fine, no names ever. I'll find the patterns and put them on future episodes. Because I'd rather tell you what is actually happening on floors than what sounds good from a stage. Buy the technology. Automate your backup house until it humps. I'm serious. Go build it. We are. Then take the hours and dollars you got back and spend them on putting a prepared human in front of a guest with enough time to notice them. That is the entire strategy. It is not sophisticated and almost nobody is doing it because the second half never survives the budget meeting. So the question I'm asking you this week, the same one I'll ask every guest who sits across from me, what did your team stop doing? When you bought that tool, and what do they do with that time now? If you cannot answer that, you already have the answer. See you next time on the return.