A 94% audience score, and they were talking about Batman
Clayface opened at $24.0 million domestic on a ~$40 million budget, number one for the weekend, with an A− CinemaScore and 94% on Rotten Tomatoes — the best audience score DC has posted. The decision that follows is what to buy with it: a sequel, or another standalone under the same umbrella. So we counted what the audience actually said.
The 94% did not attach to the film. It attached to a character who is not in it.
Across 973 comments, with ten terms registered before any data was pulled: Batman 10.28%. The DCU 2.88%. Gunn 1.34%. All five craft terms together 1.03%. A sequel 0.41%. And the word makeup appeared zero times on a film whose production story is practical makeup effects.
The decision this is for
A studio head deciding what the next mid-budget title from owned IP looks like. You have a hit on a $40 million film. Do you greenlight the sequel, or another standalone under the same umbrella?
The audience that showed up is attached to the umbrella at twenty-five times the rate it is attached to this title. Four comments in 973 asked for more of it.
What 973 people said
| Term | What it is | Comments | Share |
|---|---|---|---|
| Batman | a character who is not in the film | 100 | 10.28% |
| DCU | the universe it was built to seed | 28 | 2.88% |
| Gunn | the man running that universe | 13 | 1.34% |
| Superman | the flagship | 8 | 0.82% |
| creature | how it was made | 5 | 0.51% |
| sequel | whether there should be another | 4 | 0.41% |
| practical | how it was made | 3 | 0.31% |
| gore | how it was made | 1 | 0.10% |
| prosthetics | how it was made | 1 | 0.10% |
| makeup | how it was made | 0 | 0 of 973 |
Talk about Batman runs ten times higher than talk about the craft, and twenty-five times higher than talk about a sequel. Two further rows were measured and are not in this table because they drew different corpora — 653 and 101 comments. They are reported below with their own denominators.
The $40 million did not buy a louder conversation
Two controls were registered before the pull. A craft rate with nothing beside it is a number, not a finding.
| Film | Detail | Rate | Comments | 95% range |
|---|---|---|---|---|
| Clayface | DC Studios / Warner Bros. · ~$40M | 1.03% | 10 of 973 | 0.56–1.88% |
| Wicker | Black Bear · same weekend, same genre | 0.92% | 1 of 109 | 0.16–5.01% |
| Superman | DC Studios · same universe, not horror | 0.10% | 1 of 1,000 | 0.02–0.56% |
Clayface cannot be told apart from a horror film that cost a seventh as much. The ranges overlap almost entirely. Superman is lower but only just — its range touches Clayface’s at the edge. The point estimates alone would have hidden both facts, which is why the ranges are on the page.
What is not thin is the comparison inside the single 973-comment corpus: franchise at 15.72% against craft at 1.03%, same moment, same denominator. Those ranges do not come close to overlapping.
Critics and audiences are not arguing
Clayface sits at 76% with critics against 94% with audiences. Five words a mixed review gets written in were registered before the pull and counted across 653 comments.
| Term | Comments | Share |
|---|---|---|
| predictable | 2 | 0.306% |
| generic | 1 | 0.153% |
| derivative | 0 | 0 of 653 |
| formulaic | 0 | 0 of 653 |
| shallow | 0 | 0 of 653 |
The audience is not disputing the critics’ verdict. It has not encountered it. Those are two different facts and they lead to two different marketing decisions.
| Title | Critics | Audience | Demand |
|---|---|---|---|
| Lanterns | 91% | 67% | latest three months are its best three |
| Creature Commandos | 93% | 79% | — |
| Peacemaker | 93% | 83% | keeps 46% of its best year |
| Superman | 83% | 90% | latest three months down 87% from peak |
| Blue Beetle | 78% | 90% | keeps 9% of its best year |
| Clayface | 76% | 94% | flat in band since 2022 |
| Supergirl | 52% | 72% | — |
A searcher looking for the verdict meets the critics first: page one for clayface review runs rottentomatoes.com, yahoo.com, tldrmoviereviews.com, youtube.com, kabooooom.com, thewrap.com, dreadcentral.com. Two of seven are audience surfaces, and the first result leads with the 76%. No Reddit, no IMDb, no Letterboxd.
Demand peaked six months before release
| Month | Searches |
|---|---|
| 2024 average | 23,192 |
| December 2024 — greenlit | 49,500 |
| March 2026 | 33,100 |
| April 2026 — first trailer | 823,000 |
| May 2026 | 201,000 |
| June 2026 | 110,000 |
| July 2026 | 450,000 |
| August 2026 | 246,000 |
| September 2026 — month before release | 368,000 |
The $24 million opening was collected against a curve already 55% off a peak set at the trailer. Over the last twelve months “clayface” averaged 204,600 a month against 101,875 for “superman movie” — the mid-budget horror title out-drew the flagship two to one.
What we registered before the data was pulled
Five predictions, each labeled with what it was worth. Two came out wrong, and the one we were most confident about missed by fifteen times.
| # | Prediction | Band | Basis | Result | What happened |
|---|---|---|---|---|---|
| C1 | Craft talk out-totals franchise talk | likely | declared guess | Missed | Franchise ran 15.72% against craft at 1.03% — fifteen times the other way. Stripping Batman, which was registered in advance as inflating the franchise row, still leaves 5.3×. |
| C2 | No critic-vocabulary term clears 2% | likely | measured, 1/1 | Held | Highest single term 0.306% across 653 comments. The only prior here grounded in an earlier measurement. |
| C3 | Emotional themes clear 5% | roughly even | declared guess | Missed | Zero of 101. The rule-of-three upper bound on a zero count at that size is 2.97%, which still clears the falsifier — but 101 is not the 973-comment corpus the other rows read. |
| C4 | Search more than 10× its 2024 baseline | likely | cross-instrument, discounted | Unresolved | October is excluded by the engine as an incomplete month, so the release spike is structurally invisible. Registered in advance to resolve unresolved if so. |
| C5 | Audience surfaces hold under half of page one | roughly even | declared guess | Held | Two of seven. |
A sixth prediction, registered four days earlier, also resolved here. Clayface was called to open between $22 and $34 million domestic at roughly even odds. It opened at $24.0 million. Post-review tracking published after the call came in at $18–20 million, below the band, and the recorded response at the time was that it moved confidence and not the number. The tracking was wrong and the band held.
The full record is at the prediction ledger.
What this study cannot tell you
- The corpus re-samples per row, and that nearly cost the finding. One query string, one run, four rows — and the denominators came back 973, 973, 653 and 101. The central comparison survives only because the two rows it rests on happened to draw the same 973 comments. Every denominator on this page is printed beside its rate and no share is compared across rows that do not share one.
- Theme frequency is not sentiment. Batman at 10.28% says what people are discussing, not whether they enjoyed the film.
- A YouTube comment corpus is a commentary corpus — reactors, explainer channels, reviewers — not a viewer sample. Some of it is semi-professional criticism, which biases against the two predictions that held.
- October demand is invisible. The engine excludes the incomplete current month, so the opening-weekend spike is not in the series. A re-pull after November 1 costs $0.011.
- One of eleven rows was refused — a SERP row missing a required field in the specification we wrote. It supported no registered prediction and is formally abandoned here with its cause named, rather than quietly dropped.
- Craft talk cannot be told apart from the genre control. Clayface at 1.03% and Wicker at 0.92% have 95% ranges that overlap almost entirely. Superman is lower but only just — its range touches Clayface’s at the edge. The point estimates alone would have hidden both facts.
- Four films is four films. The critic-against-audience ordering on the slate is clean, and a perfect ordering of four happens by chance one time in twenty-four. This pass did not test it.
- No comment text exists in this output. The engine never stores it. Nothing here is quotable and none of it was retained.
How this page changes
A standing study, not an article. Corrections are published in place and dated, never applied quietly.
| Measurement | Cadence |
|---|---|
| October search — the release month | As soon as October closes, $0.011 |
| The ten registered theme terms, with denominators printed | With each re-pull |
| Both controls | With each re-pull |
| Rotten Tomatoes and CinemaScore figures | These move; re-read at each update |
| Date | What changed |
|---|---|
| 2026-10-26 | First publication. Two registered predictions came out wrong and are published above. One row of eleven was refused for a defect in our own specification and is named in the limits. The engine returned four different corpus sizes for one query string in one run, and the central finding survives only because the two rows it rests on happened to match — that is recorded as luck, not design. |
October is the month this film actually opened in, and the instrument cannot see it yet.
Sound Decisions carries the re-pull when it lands, with the registered question written before the data is.
Subscribe to Sound Decisions