Skip to the main content.
Log in Sign Up
Log in Sign Up
Facebook Instagram Linkedin X YouTube Tiktok
RESEARCH REPORT:

How accurately do humans review transcripts?


By surveying 250+ application readers in a first-of-its-kind report, we establish a baseline benchmark for human transcript reading accuracy.

Access the full report PDF:

 

Get the full report

report cover-2

 

"Not all transcripts are the same. Some put the graduation day at the very top, some put them at the very bottom"

"International transcripts, they're not the same as American transcripts"

38% of responses cited format variability & inconsistency

"Some schools have different grading systems so we had to figure out what some grades meant"

"Deciphering course codes"

27% of responses mentioned understanding complex information


Key takeaways

Screenshot 2025-08-13 at 9.41.56 PM

 

Manual transcript processing is incredibly labor-intensive 

Self-reported transcript processing times had a median of 20 minutes per transcript. Much of the work involves calibrating GPAs and translating data between systems.

Human errors are expected

Due to the vast variability in formats, as well as competing priorities in admissions offices, human transcript readers are prone to make mistakes when reading transcripts.

Leveraging AI could significantly improve accuracy and speed of reading

Where traditional OCR solutions have failed, the ability of AI to pattern match across near-infinite formats introduces new opportunity for more accurate, faster, and less laborious transcript reading.

297 higher education admissions professionals surveyed, with 252 having direct transcript reading experience. Data collected July 2025 in partnership with Centiment.