The attribution gap

Every promoter has had this conversation. The platform dashboard says the campaign returned 6.2x. The settlement says the run is four hundred seats short of where it needed to be. Both numbers come out of serious systems, and both are correct inside their own frame. Only one of them is the business.

Two ledgers, one show

An ad platform's ledger records what it can observe: a click, a pixel fire, a modelled conversion, a value handed back by a tag it does not control. A ticketing system's ledger records what happened: a seat, a price band, a fee, a timestamp. The first ledger exists to explain the platform's own contribution. The second exists to survive an audit.

When the two disagree, the instinct is to reconcile them. That is the wrong move. They are not two measurements of one thing. They are two different things, and only one of them pays the artist.

Why the gap exists

1. Modelling fills the holes

Consent rules, tracking prevention and app privacy frameworks removed a large share of deterministic signal. Platforms did not respond by reporting less. They responded by modelling more. Modelled conversions are a reasonable engineering answer to missing data. They are not a receipt.

2. The windows overlap

A seven day click and one day view window on one platform, a thirty day window on another, and a last non-direct rule in the analytics suite will each claim the same ticket. Nobody is lying. Every platform is answering the question it was asked: what did I touch? Put three platforms behind one on-sale and the claimed volume can exceed the tickets that exist.

3. The purchase completes somewhere else

In live entertainment the transaction usually finishes in a ticketing environment the promoter does not own. Sometimes on a partner domain, sometimes inside an app, often after a queue. Every handoff is a place where value gets lost, duplicated, or handed to the last thing that moved.

What the gap costs

It matters because budgets are set from it. Take a fifteen show run where the platforms together report 24,000 purchases and the ticketing system reports 17,400 sold. Nothing fraudulent has happened; the overlap is arithmetic. But if the allocation for the next three on-sales is written against the 24,000, roughly a quarter of that spend is being steered by tickets that do not exist.

The cost is rarely a visible failure. It shows up as drift: money staying too long on a channel that reports well, mid-week dates nobody flagged in time, a pricing decision taken against a demand curve that was partly synthetic.

Measure the thing that settles

The answer is not a better attribution model. It is a change in which ledger is authoritative. Once ticket level sales are the input, the useful questions become answerable.

None of this needs a new pixel. It needs the sales data and the spend data in one place, at the same grain, on the same clock.

Closing the gap

That is the job of the NYBA platform. It reads every ticket sale from the systems already in use, joins it to spend across channels, forecasts demand per show, and moves budget toward the campaigns that produce sold seats rather than reported purchases. The ticketing stack stays where it is. The reporting layer is what changes.

Across more than 1,200 events a year the pattern holds: the platform ledger and the box office ledger never converge, and the organisations that plan against the second one make better calls earlier.