Observation Report
Observations, spoken vs typed.
The Q1 2026 data.
TeachScribe is an AI-powered observation tool for Early Years teachers. Instead of typing up observations by hand, often hours after the moment has passed, teachers speak while they teach. A small wearable push-to-talk microphone captures the audio, and our AI turns it into a structured observation linked to the right child and the correct EYFS goals.
This report analyses a random sample of 3,000 real observations collected between 1 January and 31 March 2026 across schools using TeachScribe. It compares two ways of recording: AI / spoken (the teacher speaks and TeachScribe does the rest) and manual (the teacher types the observation on an iPad).
Speed comparison
Median time to record a single observation, measured on the observation screen itself.
AI / Spoken
16.5s
Median time per observation
Manual typed
3m 2s
Median time per observation
The gap
11× faster
The median spoken observation takes 11× less time than the median manual one, and that's a conservative number. The manual timing only measures time on the observation screen itself. It doesn't include picking up the iPad, logging in, or navigating to the right child. The true end-to-end gap is wider.
Volume split
Teachers choose voice 15 times out of 16.
Across the quarter, 93.8% of observations were captured by voice and 6.2% were typed manually. That's roughly 15 spoken observations for every 1 manual entry.
Both options are available to every teacher, at all times. This ratio isn't a quirk of availability. It's a strong revealed preference for the voice workflow over typing.
School hours (08:30–15:15) vs outside
Voice observations happen live. Manual ones drift into the evening.
87.6% of spoken observations happen during the school day. Manual observations tell a different story: 35.8% are recorded outside school hours, 3× the rate for voice.
| Observation type | During school hours | Outside school hours |
|---|---|---|
| AI / Spoken | 87.6% | 12.4% |
| Manual typed | 64.2% | 35.8% |
Hourly distribution
When in the day do observations actually happen?
AI / spoken observations peak at 10:00 and 14:00 with a clear lunchtime dip, matching the morning and afternoon session pattern. Manual observations spread across the day and have a long tail into the evening, with teachers still typing up observations between 8 and 10 PM.
Source: random sample of 3,000 observations from 1 Jan – 31 Mar 2026, grouped by hour in UK time.
What this means
Three takeaways from the data
Observations move from after-hours to in-the-moment.
Voice observations cluster tightly within school hours; manual ones drift into evenings. Spoken observations are captured live. Manual ones are often recalled and typed up later.
The 11× speed gap is probably understated.
The 16.5s vs 3m 2s comparison only counts time on the observation screen. It doesn't include finding the iPad, logging in, or navigating to the right child. The voice workflow skips all of that.
Given both options, teachers choose voice 15-to-1.
Both methods are always available, so the 94%/6% split reflects genuine preference, not limited access. When typing is available and voice is available, voice wins.
Want your setting in the next report?
Book a short call and we'll show you how TeachScribe works for your team, plus what the numbers look like for a setting your size.
Book a call