AI in Education · For teachers

Did your feedback pilot reach more students, or did the same students submit again?

The short answer

Pick a class, a time window and an event you can verify, such as feedback released to a student. Count each eligible student once in that window. Report the number reached alongside the number eligible. Keep repeat releases in a separate total so a busy revision week cannot disguise a gap in coverage.

Four things to remember

  1. Put the eligible class size beside the number of distinct students with a recorded release.
  2. Count a student once per reporting window, even if several revisions received feedback.
  3. Choose a label that matches what the record actually shows.
  4. Report uncertain records separately and say how they could change the result.

What would count as reaching a student in this pilot?

A head teacher has two figures for the faculty meeting. Six pieces of feedback went out on the first English task and eight on the second. The team wants to know whether more students had a chance to use feedback. Those two totals cannot answer the question. One student who redrafted a paragraph three times could account for much of the difference.

Write the question on the top of the sheet before opening the activity report. For this exercise it is “How many students in the participating class had feedback on this task released to them by Friday?” A saved teacher comment falls outside that definition. A second release to the same student counts as another release, but the student already has a place in the headcount.

Use the strongest wording your records support. If the school can verify only that a student submitted work, call the measure submissions. A release record supports “made available”. To investigate whether students read the comment or could act on it, ask a separate question. A revised passage in which a student has tried the suggested change gives the teacher something concrete to inspect.

How can eight releases reach fewer students than six?

Try this fictional faculty exercise. The same ten students were eligible for both tasks. Each letter stands for one student. Each appearance of a letter is one confirmed, separate release of feedback on a piece of work. There are no duplicate system rows or unconfirmed entries in this first example.

For the first task, the release list is A, B, B, C, D, E. For the second, it is A, A, A, B, B, C, C, D. Before reading the answer, cross out the second and later appearances of each letter in each list. Circle anyone who appears for the first time in the second week.

The first list leaves five students. The second leaves four, all of whom were in the first list. Feedback therefore reached five of ten students in week one and four of ten in week two. There was one release beyond a first release in the first week and four in the second. Those four additional releases went to three students, A, B and C. The team might be pleased to see students working through further drafts. It still needs to ask who had a first chance.

Across the fortnight, A through E each appear at least once. That is five distinct students, even though the two weekly headcounts add to nine. Use the full fortnight as the window and cross out repeated letters across both lists if the principal asks for a fortnightly figure.

Who belongs in the denominator?

Start with the group the school intended the activity to reach. For a whole-class pilot that might be the students on the agreed class list for that task. Using only the students who uploaded a response removes the very students whose access the faculty may want to investigate. Record when you took the class list and what eligibility rule you used.

For a second practice check, imagine four of eight eligible students had feedback made available on one task. On the next task five of twelve did. One more student was reached, but the proportion fell from one half to five twelfths, about 42 per cent. Both statements belong in the meeting note. With a changed cohort you would need the two lists to establish which individuals were newly reached.

The RE-AIM framework describes participation using both the people who participate and the eligible population. Its guidance comes from evaluating interventions. Our classroom exercise borrows that counting distinction. A single class percentage cannot tell a faculty whether every group had a fair chance to participate.

What should you do with a missing student or a doubtful row?

Suppose the second week in the ten-student example also includes one confirmed release addressed to an eligible student whose identifier is missing from the report. Four students are identifiable. The extra release might belong to one of them, or to a fifth student. Record four confirmed, with a possible upper count of five. Dividing nine releases by ten students gives neither of those answers.

Ask the staff member with approved access to reconcile the unresolved row against the school record. A repeated export row for the very same release should count once even in the release total. A second, genuinely released revision is a different event. Keep identifiable working records in the school’s approved location. For a training exercise or slide, the invented letters above will do.

For an AI-assisted pilot the Australian Framework for Generative AI in Schools calls for monitoring how the technology affects the school community. Use the school AI pilot checklist to assign responsibility for a follow-up. For example, a teacher could check whether students outside the online release list received paper feedback on the same task. Record the reporting scope before combining the two sources.

What can a head teacher take to the review meeting?

Copy the blank record below for the next task. For the worked example, a defensible meeting note reads, “Eight releases reached four of our ten eligible students on task two. All four had received feedback in week one. Check with the teacher how the other six had access to feedback before we extend this activity.” That gives the faculty a question it can act on without guessing why individual students were absent.

For questions about the quality of the comments, use the guide to what to measure in a school AI feedback pilot. Pair one count of access with a small review of student work. Each answers a different question for the team considering whether to expand the pilot.

If your faculty is considering JeddAI, bring the two fictional release lists to a school enquiry. Ask which stages of submitting, reviewing and releasing work your staff could verify, and which evidence they would need to collect at school. It is a more precise conversation when everybody knows which event is meant by “reached”.

Answer key for the ten-student English class
Measure Week one Week two
Eligible students 10 10
Confirmed releases 6 8
Distinct students 5 (A–E) 4 (A–D)
Share of eligible class 5/10 = 50% 4/10 = 40%
Releases beyond the first per student 1 4
Copy this record for a faculty pilot review
Field What to enter
Class, task and time window Which students had the opportunity, on what task and by when?
Eligible cohort Number eligible, source of the list and date checked.
Evidence of release Which event is counted? Which channels are covered?
Confirmed events and students Total separate releases ___; distinct eligible students ___ of ___.
Unresolved rows What is missing? How many additional students could this represent?
Decision and follow-up What can the faculty conclude? Who checks the gap, and by when?

Questions teachers ask

Should a student who received feedback twice count twice?

Count two genuine release events and one student in that reporting window. Keep both figures if you want to understand revision activity as well as access.

Can we add the weekly headcounts to report a term total?

Count distinct eligible students across the term itself. Adding weekly headcounts counts students again whenever they appear in another week.

What if the pilot covered online feedback but students also received it on paper?

Label the result as online coverage. To report on both channels, reconcile their records for the same task and period using the school’s approved process, counting each eligible student once.

Does a recorded release show that the feedback helped?

It shows that feedback was made available through the channel you checked. To investigate use, inspect what students did with a particular comment or ask them to explain their revision.

About this guide

Written by Shashank Singhal and reviewed on 11 Nov 2026. Filed under AI in Education. Looking for study material? Browse Jeddle’s Australian-English study guides.

Keep reading

All teacher guides
Shopping cart0
There are no products in the cart!