Test screenings have a reputation for producing compromised films by committee. The process is narrower than that and the changes it produces follow a consistent pattern.
The audience is recruited to a specification
Attendees are selected against the demographic the film is aimed at, then screened for whether they would plausibly buy a ticket.
They see an unfinished version, frequently with incomplete effects, temporary music and no final grading, and are told as much beforehand.
Because the version is unfinished, some reactions measure the state of the print rather than the film, and experienced analysts discount accordingly.
What the cards actually measure
Viewers rate the film overall, name scenes they liked and disliked, and state whether they would recommend it to a friend.
The recommendation question carries the most weight, because it predicts word of mouth, which is what determines performance after the opening weekend.
A smaller discussion group follows the screening, and that conversation identifies why a score was low in a way the cards cannot.
Confusion is the most actionable finding
The clearest signal a screening produces is where an audience lost track of what was happening or who someone was.
Confusion is fixable, usually through a line of dialogue, a reordered scene or a clarifying shot, and it is fixed almost every time it is found.
Dislike is a weaker signal, since a scene an audience finds uncomfortable may be doing exactly what it was built to do.
Endings change more than anything else
The ending determines how an audience feels leaving the cinema, and that feeling drives the recommendation question directly.
Reshooting an ending is also comparatively contained, requiring fewer locations and less time than restructuring a film.
This combination of high leverage and low cost is why endings are the element most often revised after a preview.
Why the process is blamed unfairly
A screening produces information rather than decisions, and what is done with the information is chosen by people with their own views.
Filmmakers with sufficient standing use the results selectively, and those without it may find the results used to justify changes already wanted.
The complaint about test screenings is therefore usually a complaint about who holds authority over the final cut, which is settled in a contract long before any audience sees anything.