How film test screenings work

By Joey Dawson, Dawson's

Last reviewed: July 2026

A test screening is a preview of an unfinished film to a recruited audience, held to measure how the film performs before it's locked and released. Viewers are selected to reflect the film's intended audience, watch the film in normal viewing conditions, then complete a questionnaire — often followed by a moderated focus group with a subset of them.

The results tell the filmmakers two distinct things: whether the film holds the people who watch it, and whether it would draw an audience in the first place (the trade calls these playability and marketability). Studios have used the practice since the 1920s and formalised it in the 1970s as marketing and distribution spending grew.

What happens at a test screening

1. Recruitment. Viewers are recruited to match the target audience — usually by genre interest and demographic. They're typically told the genre and age range sought, and often little else about the film. Secrecy levels vary with how protective the production is.

2. The introduction. A moderator explains that the film is unfinished: the sound mix, visual effects, score and titles may all be temporary. This matters — without it, audiences mark down a temp score as though it were the final one.

3. The screening. The film plays. Some companies also film the audience to capture reactions not audible from the back of the room, and at least one European research firm has built its offer around facetrack technology for this purpose.

4. The questionnaire. Immediately after, before the audience discusses it. Historically paper; now usually digital, which allows skip logic — follow-up questions triggered by earlier answers, so different segments of the audience get different questions without any of the group-think a focus group produces.

5. The focus group. Typically 15–25 viewers held back for a moderated discussion, deliberately selected across the range of responses rather than only the enthusiasts.

6. The report. Scores, comparisons against benchmark databases of similar films, verbatim comments, and identified problem points.

What gets measured

Overall rating — a five-point scale from excellent to poor. Studio research firms track the top two boxes (excellent plus very good) as a broad satisfaction measure, but the headline figure is the top box alone: the percentage rating the film Excellent. Enthusiasm predicts behaviour; approval doesn't.

Definitely recommend — the percentage who would definitely recommend it, top answer only. Often the most predictive single number in the report, because word of mouth drives independent film performance more than any other factor.

Benchmark comparison — how those scores compare to previous films of the same genre and budget. This is the real value of the large research firms: one leading firm has run over 3,000 in-theatre screenings and interviews over a million respondents a year, which is a benchmark database no one can replicate quickly.

Attention and drop-off — where interest falls away.

Comprehension — whether the story landed as intended.

Character response — who the audience connected with. Studios have added scenes for characters audiences responded to, and cut ones they didn't.

Description — how viewers describe the film to a friend. Underrated, and the closest thing to a live marketing test.

How much test screenings cost

Studio-scale in-theatre testing has historically run at roughly $10,000–$20,000 per screening — described by people who commission it as among the best money a production can spend. Studios usually budget for it as part of the production.

Independent testing costs less, because audiences are smaller and methods are leaner. Online testing costs less again.

For an independent producer the calculation is straightforward: the cost of one screening against the cost of discovering a structural problem after picture lock, or of mispricing the film at market.

In-cinema or online?

In-cinema reproduces real viewing conditions — a room full of strangers, no pausing, no phone. For comedy and horror especially, the collective response is the data. One French research firm runs two back-to-back screenings deliberately, to observe how the same film plays to different atmospheres.

Online reaches audiences across regions, which addresses a genuine criticism of in-theatre testing: a single screening in a single location tells you little about how a film performs nationally. It also allows moment-to-moment feedback. Security has improved substantially — identity verification, personalised watermarking burned into each stream, enforced NDAs.

Both are legitimate. Choose on what you need to learn.

Protecting the film

The reason most independent producers never test is fear of a leak. Standard protections:

• Phones collected and secured for in-cinema screenings
• Signed NDAs from all participants
• Identity verification for online viewers, with individual visible watermarks
• Recruitment that discloses genre only — no title, cast or synopsis
• Unpublicised venue and date

Ask any testing provider to walk you through their security before you send them a cut.

Using the results well

Testing produces information, not instructions. The most common failure isn't a bad score — it's misreading a good one.

A note about a scene is a symptom. The job is diagnosing what it points to. Audiences are frequently wrong about the solution and right about the feeling: "the ending didn't work" rarely means change the ending, and often means the setup two acts earlier didn't pay.

You will get feedback you disagree with. That's normal and useful. Filmmakers who once resented testing as studio interference have largely come to find it valuable, and the practice is now standard for independents as well as studios — but the value depends entirely on reading it as diagnosis rather than verdict.

Frequently asked

Read next

Excellent and Definitely recommend →
Film audience testing and research →

Test the film first

Dawson's runs audience testing for independent films, including films we don't represent.