The short answer
Three ads share one budget, and the strongest one starts winning most of it
Your budget splits across all three ads, not into three separate budgets. Delivery leans toward whichever ad people respond to, so over time the stronger ad quietly starts earning most of the showings while the weaker ones fall behind. That's not random, it's the system finding your winner for you. Your job is to watch it happen, then kill the ad that's losing.
What happens when your ads compete for the same budget
Your ads split one budget and compete for it, they don't each get their own. When you run three versions of a listing ad or a brand ad at once, you're not tripling what you spend, you're giving one pool of budget three ways to reach people. Hugo's delivery decides how to divide that pool based on what people actually do when they see each ad. Nothing about this needs you to set percentages or pick a favorite up front. You just let all three run.
This is true whether you're comparing two photos, two headlines, or two totally different angles on the same listing. The mechanics don't care what you changed between versions. What matters is that you have more than one live ad pulling from the same budget over the same stretch of time, so the comparison is fair.
How Hugo decides which ad keeps running
Delivery leans toward whatever people respond to, and it does this on its own. As your three ads run side by side, Hugo's system tracks which one people actually engage with, and it starts routing more of the shared budget toward that one. The ad that isn't landing keeps running too, but it gets a smaller and smaller slice. You don't have to declare a winner. Within the same window of time, the budget effectively declares it for you.
This is why a small stable of ads works like a testing machine you get for free. You didn't set up a formal test, you just launched three versions of the same listing and let delivery sort them. The stronger one earns more of the showings quietly, in the background, while you're doing everything else your day requires.
Why killing the losing ad is the whole point
Killing the loser is the actual discipline here, and it's the part most agents skip. It's easy to fall in love with the ad you personally like best, the photo you picked, the line you wrote, and let it keep running out of loyalty rather than performance. But the rotation only teaches you something if the losing ad actually leaves. If you let it limp along next to the winner, you're not testing anything anymore, you're just running two ads out of habit.
The move is simple: let all three run, watch which one delivery keeps feeding, pull the one that's starving, and put a new challenger in its place. Then do it again. That's the whole cycle, and it costs you nothing extra to run it, since you're using the same package and the same budget you'd be spending anyway. The only thing you're adding is the willingness to let a weak ad go.
Does running three ads cost three times as much?
No, the budget stays exactly the same and gets split across whatever ads you're running. Hugo's delivery watches how people respond and starts sending more of that shared budget toward the ad that's pulling its weight, and less toward the one that isn't. You're not paying more for more ads. Same money, better information.
How do you know which ad is winning?
Watch delivery, not your gut. Over the same stretch of time, the ad people respond to more starts earning a bigger share of the showings on its own, without you adjusting anything. That shift is the signal: whichever ad the system keeps feeding is the one working. Trust the pattern, not your favorite.
What's the biggest mistake agents make when running multiple ads?
Keeping the loser around. Agents get attached to the ad they personally like, the headline they wrote or the photo they picked, and let it keep running even after delivery makes clear it isn't pulling its weight. The rotation only teaches you anything if the losing ad actually leaves and gets replaced by a new challenger. Kill the ad that isn't working.