Which campaign sold you tickets, and which one only sold you clicks?

There is a structural problem in how advertising for an event gets measured, and it is not technical: the platform that charges you is the same one that tells you it worked.

Meta reports the conversions Meta attributes to itself. TikTok reports its own. Add the two together and you get more tickets than you sold. That is nobody’s bug: each one counts with its own attribution window and both are claiming the same buyer.

Why staring at the dashboard harder does not fix it

Because the platform’s dashboard cannot know what it cannot see. And what it cannot see is your sale: if your event gets bought on a ticketing platform’s domain rather than yours, the pixel stops following that person at exactly the step that matters.

What comes back is a conversion declared by the platform itself, estimated with its own model. It is useful for optimising inside that platform. It is no use for deciding how you split budget between platforms, which is the decision that actually costs money.

The one-minute test

Take your last event with advertising on two networks. Add up the conversions each dashboard reports. Compare that against the tickets you sold.

If the sum comes to more than the real total — and it almost always does — you already know you are splitting budget using two figures that overlap. The question stops being which one worked better and becomes how much of what you think you know is double counting.

What it takes, and it is not a data team

Three things, in this order:

1. That the page where the purchase happens is yours. It is the prerequisite for everything else. Without your own domain there is no measurement of your own, only what a third party hands back.
2. That the pixel follows through to the purchase, not to the click that leaves. With the sale on your domain, the purchase event is yours and you measure it.
3. That the purchase is tied to a person, not to a device. Someone who bought on a laptop in March and on a mobile in August is one person in your database — and counting them twice inflates your customer count and deflates your repeat rate at the same time.

A limit worth saying out loud

None of this ties an ad impression seen on one device to a purchase made on another days later. That needs the campaign identifier captured at the moment of purchase, and it is a separate piece. Anyone promising you that closed loop without explaining how they capture it either has it built and can show you, or is describing somebody else’s attribution model.