Nine years of trailers, scored after the fact

We took 1,658 theatrical releases from 2018 to 2026, found 3,068 creator reactions to the films themselves and 2,289 to their trailers (trailer reactions are scored for 2022 to 2026 releases so far), read over 820,000 audience comments under those videos, and ran today's scorer over all of it. Then we checked the scores against what the films did at the box office, and against the measures the industry already uses: CinemaScore, PostTrak, Metacritic and Rotten Tomatoes.

These are historical model scores, not Reactr Ratings. Nothing on this page was published at the time. Each score is the current method applied to archived videos and comments years later, so it shows what the method would have said, with hindsight about which videos exist. Published Reactr Ratings are on the rated titles list and are frozen in the ledger before a film opens.

How Reactr stacks up against CinemaScore, PostTrak, Metacritic and Rotten Tomatoes

Every measure was tested on the job a theater, a distributor or a financier cares about most: telling which films will hold their audience. The test is the second-weekend hold, the second weekend's gross as a share of the opening. For each measure we ranked the films and checked that order against the order of their holds. Agreement runs from 0, no match, to 1, the same order; bigger is better. Reactr here is the Verdict: what creators said after seeing the film, pooled with the audience comments under their videos.

On the same 259 films, Reactr leads CinemaScore, Rotten Tomatoes and Metacritic

The same 259 films with CinemaScore, Rotten Tomatoes and Metacritic. The lead is about 4 hundredths, and the intervals still overlap.

Each measure on every film it covers, nine years

Different films for each measure, so this is what each one delivers in practice, not a like-for-like test.

Add PostTrak and the sample drops to 132 films, where all five are tied

PostTrak covers the fewest films, so requiring it shrinks the shared set to 132, the ones that have all five measures. The intervals overlap: Reactr 0.10 to 0.40, CinemaScore 0.05 to 0.39. Films in parentheses.

Where each one is strong

Season and genre rows use each measure's own films, so the sets differ by column. The highlighted figure is the best in its row. Limited releases are those that opened in under 1,500 theaters (718 films). CinemaScore and PostTrak poll the same audience and agree with each other at 0.86.

All things considered

1. Metacritic. Strongest on the most recent seasons (0.36 in 2025 to 2026), broad coverage including limited releases, decades of trust. It measures critics, not audiences, and is weak on horror, action and family films.
2. Reactr. Best over nine years on its own films (0.32) and on the 259 shared films (0.30); first on horror, action, comedy and the pandemic-lockdown seasons; the only read at the trailer stage. Behind the critics on 2025 to 2026 (0.28 against 0.36 and 0.32), weaker on animation and family, narrower coverage than the critics, and no track record yet.
3. Rotten Tomatoes. The broadest coverage (67% of limited releases) and strong recent seasons (0.32). Weak on action and family.
4. CinemaScore. The industry's audience standard for decades, polling real ticket buyers on opening night. On the second weekend it sits mid-field (0.23), it fell to 0.08 when theatrical windows collapsed during pandemic lockdown, and it covers wide releases only (11% of limited releases).
5. PostTrak. The same audience as CinemaScore, with demographic detail, but the weakest on the second weekend (0.18) and the narrowest coverage.

Against CinemaScore, the decades-old audience standard, Reactr already matches or beats it: level or ahead on identical films (0.26 to 0.22), ahead over nine years of its own films (0.32 to 0.23), level or ahead in 4 of 4 season groups and 3 of 4 genre groups, and ahead on coverage of limited releases. CinemaScore leads on animation and family. The critics are the strongest competition on the second weekend, especially lately; they measure taste rather than audiences, and they say nothing before the review embargo.

When each read arrives

The trailer read is the earliest audience reading anyone has, months ahead of release, but it does not predict box office: across the 2022 to 2026 releases it covers it agreed with the second-weekend hold at −0.06 and with the final multiple of the opening at 0.04, which is to say not at all. Its value is how a trailer landed, for the people making the next one. The Verdict does carry information about the second weekend, and it arrives by Sunday night, about when CinemaScore does: its edge is that it exists for films the polls skip, not that it is earlier.

What else the nine years say

Method by Joe Morin, Robert (Bobby) Rashtian and Evan Friis, PhD. Who they are.

Every title in the study

The 733 releases with a Reactr read. Trailer: creator reactions to the trailer, scored for 2022 to 2026 releases. Audience: comments before release. Verdict: creator reactions to the film. With audience: the Verdict pooled with the comments under those videos. CinemaScore, PostTrak (positive), Rotten Tomatoes and Metacritic as published. 2nd weekend: the second weekend's gross as a share of the opening, over 100% when a film expanded. Reactr scores run 1 to 100. Click a column to sort.

Box office: domestic, from public reporting. CinemaScore, PostTrak, Rotten Tomatoes and Metacritic as recorded on each film's public Wikipedia article. Agreement is Spearman rank correlation with a bootstrap 95% interval. The data behind this page is at /api/history-study.json. No comment text is stored or shown, only counts and scores.

How the score works · Titles rated · The ledger · Rated channels