Here’s how it works: a film is nearly complete. The studio gathers random people - often recruited from shopping malls - and shows them the movie. Afterward, these people fill out cards rating their experience. Based on those cards, films get recut, endings get changed, characters get different fates.

Sometimes this saves films. Often it homogenises them into forgettable crowd-pleasers. And occasionally it creates disasters.

I Am Legend (2007)

The theatrical ending: Will Smith sacrifices himself with a grenade to save humanity. Heroic. Standard. Forgettable.

The original ending: Smith’s character realises the vampires have been trying to rescue one of their own. He’s been the monster - breaking into their homes, experimenting on them, killing them. The “I Am Legend” title finally makes sense. He’s the boogeyman in their folklore.

Test audiences didn’t like the complicated ending. They wanted a hero, not a moral reversal. So the studio cut the ending that gave the film meaning and replaced it with a grenade. The result: a competent zombie movie stripped of everything that made it interesting.

Little Shop of Horrors (1986)

The stage musical ends with Audrey and Seymour getting eaten by the plant, which then conquers the world. It’s darkly comic, fitting the material’s tone. The film version shot this ending.

Test audiences hated watching their beloved protagonists die. Negative feedback was overwhelming. So the studio shot a new ending where Audrey and Seymour survive and live happily ever after.

The director, Frank Oz, has called the original ending “brilliant” and the replacement “fake.” He’s right. But audiences got what they asked for, not what the material demanded.

World War Z (2013)

The entire third act was reshot based on test screening feedback. Originally, the film ended in Russia with a massive battle sequence and a much darker conclusion.

Test audiences found it confusing and unsatisfying. The studio brought in Damon Lindelof to write a new ending. The WHO laboratory sequence - Brad Pitt quietly walking through zombies - replaced what was apparently an expensive mess.

In this case, test audiences arguably improved the film. The original footage was reportedly chaotic. But the fix cost $20 million and delayed release by months. Sometimes the system works; it just costs a fortune.

Fatal Attraction (1987)

Glenn Close’s Alex was originally meant to commit suicide and frame Michael Douglas for murder. It was a sophisticated ending that made sense psychologically - a final manipulation from someone who’d lost control.

Test audiences wanted Alex dead at Douglas’s hands. They’d spent the film fearing her; they wanted catharsis. The studio obliged with the bathtub resurrection scene - Alex rises from the water, gets shot by the wife.

Close fought against the change and lost. The theatrical ending is more satisfying in a crude way; the original was more interesting. Box office returns suggest audiences made the financially correct choice.

First Blood (1982)

Rambo was supposed to die. The novel he’s based on ends with Colonel Trautman killing him. The filmed ending had Rambo commit suicide. It was a tragedy about how America treated Vietnam veterans.

Test audiences rejected the dark ending. Stallone pushed for Rambo to survive. The studio agreed. Rambo lives, goes to prison, and becomes available for sequels.

Four Rambo sequels later, the test audiences got their wish. Whether that’s good or bad depends on how you feel about Rambo killing approximately 500 people over subsequent films.

The Descent (2005)

The UK ending: Sarah escapes the cave, then the camera pulls back to reveal she’s still underground, hallucinating. She’s trapped forever with the creatures. It’s bleak and perfect.

The US theatrical release cut before that reveal. Sarah escapes for real. Happy ending. American audiences apparently couldn’t handle the British version’s darkness.

Director Neil Marshall was not pleased. The US ending doesn’t just change the tone; it removes the entire point. The film is about whether escape is possible or whether trauma consumes you completely. The US version answers that question wrong.

The Problem With Data

Test screening feedback is data. Studios love data. But data about enjoyment in the moment doesn’t capture lasting impact.

People often dislike challenging material initially. They want comfort, especially from entertainment. The cards say “I didn’t like the sad ending,” and studios hear “change the ending.” What they should hear is “this ending made me feel something I’m uncomfortable with.”

Discomfort isn’t failure. Sometimes it’s the point.

When Testing Works

Not all test screening changes are bad. Toy Story’s original Woody was actively mean; test audiences helped identify that the character needed warming. E.T.'s theatrical cut was refined based on screening feedback. The system can work when it’s about clarity rather than controversy.

The problem is using audience comfort as the metric. Some films should make you uncomfortable. Some endings should be unhappy. Art isn’t supposed to be optimised for immediate satisfaction.

Test audiences are useful for identifying confusion. They’re dangerous when asked to judge artistic choices.

Test Your Film Knowledge

Related Articles