How Do Casinos Make Money Giving Away Free Drinks?

There's a small light on the back of the video poker machine you're sitting at. You can't see it. It's facing the bartender. It's red, and until it turns green, the drink you just ordered is one you're paying for. The machine hasn't decided whether you're drunk. It's decided whether you've earned it. Those lights went into Caesar Resorts across Nevada in the autumn of 2016, and the odd thing about them is what they do. A casino spent real money on equipment whose entire job is to not give somebody a drink, which is a strange purchase if free booze is how you get people to gamble. So here's a business that gives away nearly half its bar on purpose and counts every glass while it does. In the year ending June of 2025, complimentary drinks made up close to 45 cents of every That's the state's own accounting, filed with the gaming control board. Line up ten cocktails on that bar, and roughly four and a half of them were never paid for, and the house wants every one of them metered. The story everybody tells about why is short. Alcohol loosens you up, you get sloppy, you bet bigger. The house cleans up. The drink is a chemical crowbar. It's the version you'll hear from a friend at the airport bar on the way home, and it has the shape of something true. The room has no windows and no clocks, so of course they'd want you swaying a little as well. It's a satisfying explanation. It has also had a rough time in the lab. In 2022, researchers publishing in the Journal Diction gathered 20 published studies that had all done roughly the same thing. Dot give people alcohol control drinks, sit them at a gambling task, and watch what happens. Individually, those studies had gone both ways. Some found people betting more. Some found nothing. Pulled together, the overall effect on risk-taking disappeared, and there's a detail in that analysis that's worth the whole argument. Where an effect did turn up, it turned up against people who knew they hadn't been drinking. Compare drinkers against a placebo. Instead, people who believed they'd had alcohol but hadn't, dot, and it vanished. What moved the behavior was the expectation, not the ethanol. Which brings us back to the light, because now it makes sense. Nobody buys a machine to slow down their best product. Nobody involved put it that way. To be fair, operators called it consistency and better service. One long-time Las Vegas analyst thought it was mostly about bartenders comping cash customers and pocketing the difference, and calling it rationing is my inference rather than anyone's stated purpose. But every one of those explanations ends in the same place with the drink wired to a meter. Picture the seat again. Somebody is two hours in, tapping a button, half-watching a game on a screen above the bottles. A drink arrives. Nobody asked how much they were down. So what is the light reading? Not your blood, your money, and more specifically, the speed of it. Sit down and feed a 20 into a machine. From that second, the casino is measuring one thing above all others, and it isn't what you win or what you lose. It's coin-in. The running total of everything you've pushed through the machine, including the sameness money going around again and again. Win a handful of credits and push them straight back in, and the meter counts that too. Your winnings are just fresh wagering. The money doesn't have to be new. It only has to keep moving. That distinction is the whole trick. On the Las Vegas Strip, penny slot machines kept just under 11 cents of every dollar wagered through them over the 12 months ending in October of 2025. Again, that's the state's figure, not somebody's guess. So your $20 isn't $50 to them. It's a 20 that might go around five times or 50. Every hundred dollars spent is worth about $10.70 to the house in expectation. Which is why the drink isn't a generous gesture. It's small change out of a number the machine is generating, whether you notice it or not. And the way that number turns into a drink is written down in the least glamorous place imaginable. Patent filings. A 2001 application from Shuffle Master, a company that builds card shufflers and table tracking systems, lays the chain out in flat, bloodless language. Take a player's total wagering. Apply the house advantage. What you get is a theoretical win. What the Casanova expects to take from you mathematically, whatever actually happens on the night. Then apply a comp factor, a percentage the house sets for itself, and that's what comes back to you as complimentary things. Rooms, buffets, the glass at your elbow. At a blackjack table, that first number has to be guessed at. Chips change hands. Nothing gets inserted anywhere. So the pit estimates your average bet, multiplies by how long you sat there, multiplies by hands per hour, and hopes. The patents say so plainly. At the machine you're actually sitting at, nobody has to estimate anything. It counts. Which produces the part that surprises people most. The Casinos doesn't care whether you're winning. Another patent, this one filed in 2003 by Station Casinos, says it plainly. Two people playing the same game and wagering the same amount earn the same comps, even if one walks out several thousand dollars up and the other is cleaned out. The reward is priced off expected loss, not actual loss, and the same document notes, with no embarrassment at all, that big winners often get comped generously for a specific reason. To keep them in the building long enough for the house to win some of it back. So green never meant the bartender liked you. Green meant the meter said yes. Red meant not yet. When Caesars confirmed it had switched the system on across its Nevada resorts, the company's own explanation was about play, not people. The system let them serve complimentary drinks to the players choosing max play at the video poker bar tops. For decades that decision had belonged to a person. A bartender watched you, sized you up, made a call. A warm smile or a generous tip could tilt it. Now a small plastic light makes the call and it is significantly harder to charm. And the thing being rationed is almost weightless. Across those same Nevada casinos, the cost of the goods behind the bar came to about $0.15 for every dollar of drink served. A cocktail the property values at $20 holds roughly $3 of anything you could actually pour. The glass was never the expensive part, the seat was. Here's where I'll stop reporting and tell you what I think. That little light is the most honest object on a casino floor. For most of the history of Las Vegas, the free drink pretended to be hospitality. A favor, a bit of generosity, the house being nice to you, the light doesn't pretend. It states a price and it shows you whether you've met it. When those systems went in, plenty of regulars were furious and I understand the reaction. But I don't think what bothered people was that the deal got worse. I think it was that the deal got legible and that's a different thing to be angry about. Once you can see the shape of it, it turns up in places with far less neon. The coffee app that hands you a free drink at a certain level of spend. The supermarket card that prices your milk differently, depending on whether you let it watch you. The hotel tier that upgrades you for reasons that have nothing to do with how tired you look. None of those are casinos and the stakes aren't remotely the same. But they're all running the same instrument. Somewhere a number is quietly accumulating, a threshold sits across and something lands at your elbow at no charge. Which was the one moment the system tells you, out loud, exactly what you're worth to it. The gift is the readout. It's the only part of the machinery you're ever allowed to see. So the next time something is free, the interesting question probably isn't what it costs them. It's what it was measuring.