Box office bombs like The Thing (1982) and Fight Club (1999) lost millions initially. They gained cult status through home video sales and streaming.
Data analysis by DropThe
We Analyzed 207,864 Movies. These Lost Millions and Still Won.
The film industry has a dirty secret: some of the best movies ever made were commercial disasters. We pulled budget and revenue data from our database of 207,864 films and cross-referenced it with audience ratings. The pattern is unmistakable — critical acclaim and box office success are barely correlated.
Out of every movie in our database with a budget over $20 million, a rating above 7.0, and more than 1,000 votes, 142 films lost money. Not “broke even.” Lost money. Some lost nine figures.
Here are the ones that flopped hardest and aged best.
The Biggest Budget Bombs With the Best Ratings
The Irishman cost $159 million to make. It earned less than $1 million in theaters. Netflix bought it, so Scorsese did not care about box office. But by any traditional metric, this is a catastrophic return: 0.6% of budget recovered. Audience rating: 7.6 out of 10 across 7,342 votes.
Luca had a $200 million budget and pulled in $51 million — a 74% loss. Pixar sent it straight to Disney+. It holds a 7.8 rating from 8,814 voters. Another streaming sacrifice.
The Iron Giant is the textbook case. Warner Bros spent $50 million on it in 1999 and earned $23 million back. The marketing was botched. Brad Bird nearly left animation. Twenty-five years later, it sits at 7.9 with over 6,000 votes and gets cited as one of the greatest animated films ever made.
Society of the Snow cost $65.5 million. Theatrical revenue: $1,281. Not $1.2 million. One thousand two hundred eighty-one dollars. Rating: 8.0 from 3,400 votes. Netflix bought it, nominated it for Oscars, and the box office number became a footnote.
The Pattern: Streaming Killed the Box Office Metric
Six of the top 20 highest-rated box office bombs went directly to streaming platforms. The economics changed but the measurement did not. Calling Don not Look Up a flop ($75 million budget, near-zero theatrical) ignores that 152 million Netflix households watched it in the first 28 days.
| Film | Year | Budget | Box Office | ROI | Rating | Votes |
|---|---|---|---|---|---|---|
| The Irishman | 2019 | $159M | $0.9M | 0.6% | 7.6 | 7,342 |
| Luca | 2021 | $200M | $51M | 26% | 7.8 | 8,814 |
| The Iron Giant | 1999 | $50M | $23M | 47% | 7.9 | 6,028 |
| Society of the Snow | 2023 | $65M | $1.2K | 0.0% | 8.0 | 3,400 |
| The Last Duel | 2021 | $100M | $30M | 31% | 7.4 | 3,745 |
| Don’t Look Up | 2021 | $75M | $0.8M | 1% | 7.1 | 9,089 |
| Turning Red | 2022 | $175M | $21M | 12% | 7.3 | 5,671 |
| The Tomorrow War | 2021 | $200M | $14M | 7% | 7.5 | 3,966 |
| Mr. Nobody | 2009 | $47M | $2.3M | 5% | 7.8 | 6,114 |
| Gattaca | 1997 | $36M | $12.5M | 35% | 7.6 | 6,769 |
The Cult Classic Pipeline
There is a predictable path from bomb to beloved. It takes 5-15 years and requires exactly three conditions: the film has to be genuinely good, it has to be discoverable (home video, then DVD, then streaming), and it has to find its tribe.
Brazil (1985) made $9.9 million against a $15 million budget. Terry Gilliam fought Universal Studios so viciously over the final cut that the battle became its own documentary. Today it holds a 7.7 rating. It is studied in film schools.
Gattaca earned $12.5 million on a $36 million budget in 1997. Critics liked it. Nobody saw it. It became a staple of science fiction discourse over the next decade. Now at 7.6 with nearly 7,000 votes, it outrates most sci-fi films from that era.
The King of Comedy (1982) earned $2.5 million on a $19 million budget. De Niro and Scorsese made a film about obsessive fandom that audiences did not want to watch. Forty years later, Todd Phillips cited it as the primary inspiration for Joker, which made over $1 billion. The King of Comedy holds a 7.8 rating.
Mr. Nobody is the most extreme case in recent memory. Budget: $47 million. Theatrical revenue: $2.3 million. A 95% loss. Jared Leto starred before anyone cared about him. The film sat unwatched for years until it hit streaming platforms and exploded. Current rating: 7.8 from 6,114 votes.
Once Upon a Time in America: The Ultimate Bomb-to-Classic
Once Upon a Time in America (1984) is the crown jewel of this phenomenon. Sergio Leone spent a decade making it. The studio cut it from 229 minutes to 139, rearranged scenes chronologically (destroying the entire narrative structure), and released it to universal confusion. Budget: $30 million. Revenue: $5.5 million.
The restored version eventually reached audiences. Rating today: 8.4 from 5,886 votes. That makes it one of the highest-rated films in our entire database of 207,864 movies. It took 40 years for the market to catch up to what Leone built.
What the Data Actually Shows
We ran the numbers across our full database. Of the 3,247 films with budgets over $20 million and more than 1,000 audience votes:
- Average rating of profitable films: 6.4
- Average rating of films that lost money: 6.7
- Films rated above 7.5 are 2.3x more likely to have lost money at the box office than films rated between 5.0 and 6.0
The correlation between box office performance and audience quality ratings is -0.04. Effectively zero. Making money and making something good are statistically unrelated.
The best predictor of box office success is not quality. It is marketing spend, franchise recognition, and release timing. The best predictor of lasting cultural impact is the rating itself.
The New Economics
Streaming changed the math permanently. A film like Society of the Snow — $1,281 in theaters, 8.0 rating — would have been labeled a disaster in 2005. In 2023, it was a content acquisition that drove subscriptions. The metric moved from tickets to attention.
But the old theatrical bombs proved something the industry already knew: audiences are unreliable predictors of what they want until they have it. The Iron Giant, Brazil, Gattaca, The King of Comedy — none of them found their audience opening weekend. All of them found it eventually.
The data says: if it is good, it will find people. The only variable is time.
Sources: DropThe Entity Database (207,864 films analyzed), Box Office Mojo historical data, TMDb audience ratings