Quick start
Every file is a static download over HTTPS. JSON files allow cross-origin requests (Access-Control-Allow-Origin: *), so a web page can fetch them directly. The catalog lists every dataset with its fields, row count and links:
curl https://www.markstudios.com/data/index.json
curl https://www.markstudios.com/data/caption-reading-speed.jsonEach JSON file has the same shape: name, description, about (where the numbers come from), source_page, license, attribution, updated, fields, count and rows, an array of objects keyed by field. Numbers are numbers; a blank cell is null.
const res = await fetch('https://www.markstudios.com/data/video-editing-benchmarks.json');
const { rows } = await res.json();
console.log(rows.filter((r) => r.table.startsWith('Scheduled delivery window')));Video editing outsourcing benchmarks
Aggregates from Mark Studios' own order records, Jan 2023 to Sep 2026: delivery windows, what clients add to a long-form edit, raw footage per order, and repeat-client rates. One row per chart bar or table cell, with its unit and the number of orders behind it. From Video Editing Outsourcing Benchmarks, where the method and every chart are.
- JSON: https://www.markstudios.com/data/video-editing-benchmarks.json (76 rows)
- CSV: https://www.markstudios.com/research/video-editing-outsourcing-benchmarks/benchmarks.csv
- Fields:
table,row,value,unit,n - Last changed:
What buyers ask before hiring a video editor
What 698 buyers said in their first message to a video editor (Jul 9 to Sep 22, 2026): what they want edited, what they send, the questions they ask before ordering. Shares and counts per category; cells under 10 threads are suppressed. From What Buyers Ask Before Hiring a Video Editor, where the method and every chart are.
- JSON: https://www.markstudios.com/data/buyer-questions.json (148 rows)
- CSV: https://www.markstudios.com/research/what-buyers-ask-before-hiring-a-video-editor/buyer-questions.csv
- Fields:
table,row,share_pct,count,base,note - Last changed:
Caption reading speed, per video
Per-video results for 387 public videos (139 NASA, 248 MIT OpenCourseWare lectures), 339,521 captions measured with our caption checker against Netflix, BBC and DCMP limits: words a minute, characters a second, line length, time on screen and gaps, with a link to each caption file. From Caption Reading Speed Study, where the method and every chart are.
- JSON: https://www.markstudios.com/data/caption-reading-speed.json (387 rows)
- CSV: https://www.markstudios.com/research/caption-reading-speed-study/videos.csv
- Fields:
source,group,title_or_course,date,captions,minutes,words_per_minute_overall,median_chars_per_second,pct_over_20cps,pct_over_160wpm,pct_line_over_42,pct_under_5_6s,pct_over_7s,pct_gap_under_2_frames,longest_line,caption_file_url - Last changed:
The caption files themselves belong to NASA and MIT OpenCourseWare and stay under their own terms; the dataset holds only our measurements and links to the files.
License and citation
All three datasets are free to reuse, including commercially, under CC BY 4.0: credit Mark Studios and link the dataset's research page. For example:
Mark Studios. "Caption Reading Speed Study." https://www.markstudios.com/research/caption-reading-speed-study
We publish aggregates only. The order and buyer datasets hold no client names, usernames, order numbers, prices or revenue, and no cell is built from fewer than ten orders or threads. The JSON files are rebuilt from the CSVs whenever a research page is updated. Found an error? Email contact@markstudios.com and we'll fix it and say so on the page.
Tools that use the same rules
The caption data was measured with our free caption checker, which is also open source as a command-line tool, library and MCP server: caption-checker on GitHub. More free calculators are on the tools page, and every statistics page is on the research hub.