Do you know today whether Friday will sell out?

Almost everyone knows how many tickets they have sold. Very few people know whether that is few, and that is the difference between finding out on Monday and finding out on Thursday.

The figure on its own means nothing

Three hundred tickets ten days out could be an excellent season or a serious problem. Without a point of comparison that number carries no information: it is data without context, which is why the decision ends up being made on instinct even with a dashboard open.

What turns a total into a signal is comparing it against the same point on comparable events. Not against the previous event as such — an August Thursday looks nothing like a November Saturday — but against the ones that resemble it: same day of the week, same time of year, similar kind of line-up.

How to do it yourself, with what you already have

You need nothing new to get the first version of this. You need a spreadsheet and an afternoon:

1. Take your last ten or fifteen comparable events and, for each one, tickets sold at 14, 10, 7, 3 and 1 day out. Your ticketing platform gives you that if you export sales with their date.
2. Work out the median at each cut. The median and not the mean: one event that went through the roof skews the mean and leaves you with a bar that represents nothing.
3. Lay the coming event on top. There is your curve and your gap.

The first time anyone does this, the same thing usually turns up: half the events that ended up behind were already behind ten days out, and nobody looked.

The point that really matters is not the total: it is the slope

Here is the detail that decides whether this is any use. If you watch the running total, the alarm fires late, because a total takes time to separate enough to be obvious.

What separates earlier is the rate: tickets per day. A curve losing slope on day twelve does not show in the running total yet, and it is already the signal. Watching the slope instead of the total is what moves the warning from Thursday to Monday — and those four days are exactly what the useful levers need.

From warning to forecast

What you have just built in the spreadsheet already warns you. The next step is forecasting: not just “we are behind” but “we are going to finish around here”.

The version almost everyone does by hand is to draw a line: we are tracking like the comparable, so we will finish like the comparable. It works surprisingly well and it is infinitely better than nothing. There are three things about it that can be improved, and all three matter:

A number is not a forecast; a band is. The useful question is not “how much will I sell” but “between how much and how much”. You work out two scenarios — that the current rate holds, and that demand behaves the way it did last year — and what you draw is the band between them. A band reads as what it is, an estimate; a loose number invites you to treat it as a fact.

A short signal does not get stretched. If you have been selling above the line for four weeks, that says something about the next four weeks, not about the next six months. Projecting one good month across a whole season is exactly how a forecast ends up making a fool of itself. What is correct is for the recent signal to inform the short term and for the longer view to fall back on historical behaviour.

And if there is no comparable, there is no forecast. When there are not enough similar events with enough volume, the honest answer is to draw nothing. A forecast worked out on three events looks exactly as firm as one worked out on fifty, and that is where people get burned by these tools.

That is the criterion that separates a useful forecast from a decorative one: it has to know when to keep quiet.

And why this does not come as standard with your ticketing

Because a ticketing platform is built to process transactions, not to compare today’s against those of a year ago. What that comparison needs is the full history, cleaned and with events classified so it can say which ones are comparable — and that is a data problem, not a ticket-selling one.

It is the reason none of them in this sector do it: they are not missing a screen, they are missing the organised history underneath.