Most advertising problems are open-ended. You can always acquire one more customer, ship one more unit, open one more market. Live entertainment is not built that way. A room holds what it holds, the date is printed on the ticket, and the window in which advertising can move demand closes the moment the doors open. That single constraint, a fixed ceiling meeting a fixed deadline, quietly breaks the metric most teams trust the most.

Across one touring program we ran last season in the United States and Canada, the numbers looked healthy from a distance: 664,436 tickets sold, roughly $33.7M in total revenue against about $4.0M in media, a blended return of 8.4x and a blended cost of just over $6 a ticket. Those are the figures that get quoted in a deck. They are also the figures that hide almost every decision that actually mattered.

A high ROAS often measures demand, not decisions

Return on ad spend is a ratio of attributed revenue to spend. It answers the question “did the money that ran come back?” It does not answer the question a promoter actually needs answered: “how many of these tickets would have sold without me?” Those are different questions, and on a finite-capacity event the gap between them is where the profit leaks.

One run last season sold 10,045 tickets at a cost of $2.88 per ticket, a 17x return. Read casually, a 17x line item says “pour in more budget.” Read against the room, it more likely says the opposite. A show selling at that velocity is close to its ceiling. The next dollar is not opening new demand, it is paying to reach people who were already going to buy. The return stays high on the report precisely because the audience was already there. High ROAS was a symptom of strong organic demand, not proof that the advertising was doing the heavy lifting.

Now hold that against a run in the same week: 1,514 tickets, a cost of $8.41 per ticket, a return under 6x. The instinct is to move money away from the weaker line. But the weaker line was nowhere near its ceiling, which means incremental spend there had somewhere to go. The stronger line had almost none left. The metric ranked them in exactly the wrong order for the decision at hand.

Attributed is not incremental

The distinction that governs a finite-capacity event is between attributed revenue and incremental revenue. Attributed revenue is what the platform hands back to the campaign that touched it last. Incremental revenue is what would not have happened otherwise. Branded search, retargeting and prompts served to people already deep in the buying decision inflate the first and tell you almost nothing about the second. On a show approaching sell-out, a large share of attributed sales are simply demand you would have captured anyway, wearing a campaign’s name tag.

This is why a strong ROAS can coexist with a bad budget. If the return is high because the room was always going to fill, then the money spent to “help” it fill produced very little that would not have happened for free. The report looks like a win. The margin says otherwise.

The four inputs that should drive the spend decision

On a fixed-capacity, fixed-date event, the decision to keep spending, hold or cut should not run off the return alone. It runs off four things at once, and any one of them can override the ratio.

Run those four together and the same 8x return points in opposite directions depending on context. On a weak room with real capacity left, days on the clock and a stable marginal cost, an 8x return is an invitation to keep going. On a room pacing to sell out, with the marginal cost of a ticket creeping up, that same 8x is a reason to pull budget and move it somewhere it can still change the outcome.

The logic inverts between a weak event and a near sell-out

The weak markets from last season make the point from the other side. A handful of runs closed under 1,000 tickets: one at 958 tickets and a $17.10 cost per ticket, another at 931 tickets and $14.77, a third at 663 tickets and $14.80. Their returns still printed as positive, between roughly 2.8x and 3.8x. Nothing about those campaigns was broken in the usual sense. The audience simply was not there to be bought, and no amount of ROAS optimization changes a room the market was never going to fill. Spending harder against those rooms was the real waste, and the return figure was too blunt to say so.

So the rule we work to is uncomfortable for anyone raised on a single number. On a finite-capacity event, a very high return frequently means “you have less room to add value than it looks,” and a mediocre return can mean “there is still demand here worth paying for.” The ratio is an input. The seat count and the clock decide what it means.

How we separate the two in practice

None of this is theoretical for us, and none of it requires a perfect attribution model. Working across more than 180 city runs in a single season gives something more useful than any one dashboard: a baseline for how a given format tends to pace in a given kind of market, at a given price, at a given point before the date. Once you can say what “normal” pacing looks like, you can see when a room is running ahead of it on its own. Those are the rooms where advertising is mostly along for the ride, and where the marginal dollar should be the first to leave.

The other practical test is to move the spend and watch the velocity, not the ratio. When we lift budget on a room and sales velocity does not respond, while the cost of the next ticket climbs, that is the market telling us the willing audience is thinning out. When we ease budget on a room pacing to sell out and the velocity holds, that is the market telling us the demand was never ours to claim. Neither signal shows up if you only read the blended return at the end. Both show up if you watch capacity, pace and marginal cost together, week over week, while there is still time to act on them.

A room that fills itself does not need to be paid to fill. The hardest discipline in event marketing is knowing when a great-looking number is telling you to stop.