Section five

What next

What we could not answer, and what we would like to work on with you.

What we still don't know

This research answered some questions and opened others. Four are worth naming, because they are the ones we would most like help with.

What a click to a cinema is actually worth. One distributor has built the whole model, cost per click, share of the ticket price, break-even at roughly one sale per seventy clicks, and cannot close it, because the conversion rate is the single figure nobody has. Everybody guesses at it separately.

Whether being answerable changes anything. We can show that audiences are asking, and that most films have no reliable answer waiting. We cannot yet show that fixing it brings anyone to a cinema.

Whether an owned audience converts better than a rented one. The logic is sound and the distributors who have built one are convinced. Nobody has measured it against a comparable release.

Whether the cost-of-signal rule holds everywhere. It was consistent across the markets in this research. Six sessions is not a proof.

Two things we would like to work on

A clearer picture of the handoff

The gap between interest and a ticket is not a technical problem. Exhibitors and ticketing platforms hold the last piece and monetise it themselves, and no individual distributor has the leverage to change that.

But the picture is not blank. Distributors can see traffic, showtime checks and click-throughs. Some exhibitors already share more than is generally realised, in return for advertising spend or simply because they were asked. One distributor in this research approached a large independent cinema in his market and got a straight answer: across his three most recent releases with them, 15 to 17% of arrivals bought. In Norway, an initiative led by Film & Kino has had distributors and cinemas sharing release data directly for more than a year.

What we are proposing is a working group with two approaches.

The first is to see how far the existing pieces can be assembled. What each participant can already see, what different measurement approaches have in common, and what a combined view actually shows. This is not attribution and will not become attribution. It is a question of whether enough partial views add up to something more useful than any of them alone.

The second is to consider if and how the sector could approach cinemas collectively. Distributors currently ask individually, inconsistently, and for different things. A shared, modest, well-defined ask, developed by distributors rather than by us, might be harder to refuse and easier for an exhibitor to agree to. Whether that is true, and whether it is worth attempting, is one of the things the group would decide.

The output is something the sector can use, with participants credited as contributors rather than as sources.

Practical AI for distributors

Much of the industry conversation around AI has focused, understandably, on generation: scripts, images, synthetic performance and authorship.

But that has left less room for a different, more practical conversation. Distributors are, on the whole, not using these tools for the unglamorous work they are actually good at: reading campaign data, sorting audience feedback, mapping communities, reducing the reporting burden on teams of two. Younger people in the industry are asking where to learn and finding no obvious answer.

We would like to run a small programme of practical sessions, what a distributor or film marketer can actually do with these tools today, with worked examples rather than principles. As much listening as teaching: where the useful applications are, where people are resistant, and where the line should sit.

How to take part

The working group. We are looking for a small number of distributors willing to give a few hours across a few months. Not a data submission: a conversation about what each of you can already see, what you have managed to get from exhibitor partners, and whether the numbers we have hold up against yours. Everything discussed stays unattributed unless you say otherwise, and contributors are credited in whatever gets published.

The AI sessions. Tell us what would be useful. We would rather run three sessions that answer real questions than a series that covers the field. If you have already found something that works, we would like to hear that too.

To register interest, or to tell us where you think we have got this wrong, write to rfm@usheru.com.