Original data · Updated 31 August 2026
How many revisions a video ad takes: 999 projects, cut by cut
Agencies publish promises. This page publishes our archive: 999 client projects, 9,721 finished video files, and the exact number of cuts it took to get every one of them approved. Computed from the delivery folders themselves on 31 August 2026, methodology at the bottom, tail included.
How many cuts until a client says yes
Every Filmito delivery is versioned: 1.0 is the first cut, 2.0 the first revision, and so on. Across the 742 projects that follow this convention, this is the exact distribution of where "approved" happened.
Two-thirds of all projects finish by cut three: the first cut plus the two revision rounds included with every Launch ad. That number is why the policy exists. And the tail is real: 1 in 28 projects takes nine cuts or more. Those films shipped too, because the meter is not the point. The film is.
Every number, in one list
- 19.0% of projects approve the very first cut.
- 45.6% are approved by the second cut.
- 66.2% are approved by the third cut, which is the first cut plus the two revision rounds included with every Launch ad.
- The median project is approved on cut 3.0.
- 12.4% of projects take exactly four cuts; 8.2% take five.
- 1 in 28 projects (3.6%) takes nine cuts or more. We cut them anyway, because the film ships when it is right, not when the meter runs out.
- The average project leaves with 9.7 finished files: the master plus platform formats, aspect ratios, and cutdowns.
- 454 delivered files are vertical or short-form editions built for Reels, Shorts, and TikTok.
- 33 projects shipped in more than one language, from Spanish and German to Arabic.
- 742 of 999 projects follow the studio's versioned-cut naming convention and form the revision sample; the rest are excluded, not guessed.
Cite freely with a link (CC BY 4.0). If a number here ever disagrees with a number elsewhere on this site, this page is the computed source of truth and the other page is wrong; tell us and we will fix it.
Methodology, including what we refused to publish
On 31 August 2026 we scanned the studio's complete delivery archive by filename and file metadata. A project qualified if its folder contained at least one video file; 999 did. Cut counts come from the versioned naming convention used in delivery (1.0, 2.0, 3.0 ...): 742 projects follow it and form the revision sample. Projects without versioned names are excluded from revision statistics rather than estimated. Working files and duplicate copies were ignored.
What is deliberately missing: a year-by-year timeline. File timestamps before the studio's 2021 storage migration are unreliable, and we do not publish numbers we cannot verify. The archive spans work from 2017 to 2026; the per-year split does not survive our own quality bar, so it is not here.
Become data point 1,000.
Two in three projects in the revision sample were approved by the third cut. The other third took longer. The sample describes past deliveries, not a guarantee for your ad.