Product · Causal inferenceIn development

What an exhibition actually changed, not what happened at the same time.

Attendance rises in summer, in good weather, in school breaks and when a new exhibition opens, often all at once. Curatum is building causal estimates that separate these effects: what an exhibition, a price change or a campaign did to attendance, compared with what would have happened without it.

This capability is being built. The page describes what it will do; pilot institutions help shape it.

Causal inference
SeasonWeatherSchool breaksNew exhibitionAttendance

Illustrative, with invented numbers

Effect, not coincidence

What it answers

01

Did the new exhibition bring more visitors?

Compared with what the same weeks would have looked like without it, given the season, the weather and the venue's own trend.

02

What did the price change do?

Its effect on tickets and revenue, separated from the season in which it happened.

03

Was the campaign worth it?

Whether demand moved beyond what the calendar alone explains, with an honest margin of error.

How it works

  1. 1

    Identify interventions

    Exhibitions, price changes and closures found in the data are already listed in the Studio, with whether there is enough data to study each one.

  2. 2

    Model the counterfactual

    Each venue is compared with venues where nothing changed, and with its own calendar, weather and trend, to estimate what would have happened without the intervention.

  3. 3

    Report with margins

    Each estimate comes with its interval and the assumptions behind it, and coincidences stay labelled as coincidences until then.

What you get

  • Estimates of the effect of exhibitions, prices and campaigns on attendance and revenue.
  • Intervals and stated assumptions for each estimate.
  • Interventions identified in the data, with whether each can be studied (available now).
  • Readings that keep coincidence and cause apart (available now).

Frequently asked questions

Is this available today?

Identifying interventions in the data is. The causal estimates are in development with the pilot institutions.

How much data does it need?

Daily ticketing over at least a year helps separate season from effect; more venues and more interventions make the estimates sharper.

Why not just compare before and after?

Because before and after also differ in season, weather and holidays. A causal estimate compares with what would have happened without the change.

See what your data says about your venue.

A thirty-minute call with your own questions, or explore the product demo with a fictional network first.

Explore the demo