A codec is rarely the job. A file-size ceiling is.
We reviewed 424 production video compression jobs created between September 5 and October 16, 2026. The strongest signal was not a codec or resolution preference: 306 jobs (72.2%) used an exact target size. Only three chose resolution as the primary mode.
What the snapshot showed
- The median requested target among target-size jobs was 20 MB.
- 104 target-size jobs requested an output between 10 and 20 MB.
- 401 jobs completed, 20 failed, and three were still processing.
- That makes the settled-job success rate 401 / 421, or 95.2%.
- MP4-to-MP4 accounted for 357 jobs; MOV-to-MP4 accounted for another 52.
The practical product lesson was simple: when somebody has a hard upload limit, “balanced quality” is not a useful answer. They need the tool to start from the number they must clear.
The tempting claim we did not make
Aggregate input volume was 80.6 GB and aggregate output volume was 22.8 GB, a 71.7% reduction.
It would be easy to turn that into “videos get 71.7% smaller.” It would also be wrong.
That aggregate combines different durations, codecs, resolutions, bitrates, content types, and settings. A screen recording with static frames can compress very differently from noisy camera footage at the same encoder setting. The aggregate describes the observed batch; it does not predict the next file.
Why publish the denominator
“401 jobs completed” sounds good but hides unfinished work. We reported 401 of 421 settled jobs and kept the three processing jobs visible but outside the success denominator.
The same rule applies to every result in the study: attach the observation window, scope, denominator, method, and limitation close to the number.
Reusable data
The full study includes the method, a crawlable chart and table, privacy boundary, limitations, and a follow-up experiment design whose result has not been invented.
The aggregate snapshot is also available as JSON and CSV. Each row includes its unit, denominator, scope, observation dates, and a note explaining how it can be interpreted.
The data is free to quote and reuse with attribution. It contains aggregate counts only: no videos, file names, user identities, IP addresses, or job identifiers.
Limits of the study
This is one product's observational snapshot, not a representative survey of everyone who compresses video online. Jobs are not unique people, acquisition influenced the sample, and a requested 20 MB target does not prove that every source can reach 20 MB at acceptable quality.
Those limits do not make the data useless. They make the useful claim narrower: in this observed product sample, exact file-size constraints dominated the compression choices users made.
