Workload & Efficiency
How Much Time Do Lecturers Spend
Marking? The Numbers Are Worse Than You Think
Most institutions have a rough sense that marking takes a lot of time.
Few have looked carefully at exactly how much, or what that time costs
in terms of the feedback quality students actually receive. The arithmetic
is striking.
Eduface
·
9 min read
·
Lecturers & DVCs
It is late on a Sunday evening. There are 34 essays left in the queue. You have been
marking since Friday. The comments you are writing now are noticeably shorter than
the ones you wrote on Friday: not because the work is simpler, but because your
capacity to write detailed, specific feedback has a ceiling, and you passed it some time
ago. This is not a failure of professional commitment. Research by the University and
College Union found that 26 per cent of higher education staff already work more than
50 hours per week during term time.
1
The marking workload is a structural problem,
and it has a direct cost that institutions rarely measure: the quality of feedback students
receive.
How much time do lecturers spend marking in higher education?
A typical lecturer in UK higher education with two modules and 80 students per
module will face 160 or more assessed submissions per semester. At a conservative
estimate of 20 minutes per submission for meaningful written feedback, that is over
53 hours of marking per semester, on top of teaching preparation, research, and
administration. The UCU's 2016 workload survey found that 26 per cent of higher
education staff already work more than 50 hours per week during term time.
Marking at this volume is a primary driver of that figure.
What does the marking workload look like at module level?
The easiest way to understand the marking burden is to calculate it at module level.
The numbers follow directly from standard cohort sizes and assessment frequencies,
and they do not require any additional assumptions about how efficiently marking is
done.
Marking hours per semester: the arithmetic
COHORT SIZE
ASSESSMENTS
MIN PER SUBMISSION
TOTAL HOURS
40 students
× 2
× 20 min
27 hrs
80 students
× 2
× 20 min
53 hrs
120 students
× 3
× 20 min
120 hrs
Per module, per semester. Most lecturers teach two or more modules simultaneously.
These calculations assume 20 minutes per submission a conservative figure for the kind of specific, criterion-
referenced written feedback that research identifies as most useful for student learning.
For a lecturer running two modules of 80 students each, the total exceeds 100 hours of
marking per semester before research, teaching preparation, or administrative duties
are counted. Spread across a 15-week semester, that is roughly 7 additional hours of
marking every week. The UCU (2016) found that 75 per cent of higher education staff
describe their job as stressful, and unrealistic workload was cited as the leading
cause.
1
Marking at this volume, without protected time in most contracts, is a direct
contributor to that finding.
What happens to feedback quality when marking time runs out?
The response to sustained marking pressure is almost always the same: comments get
shorter. Not because lecturers do not value feedback, but because shorter comments
are the only variable that can be adjusted when time is fixed and the submission queue
is not shrinking.
The research literature on this is consistent. Hattie and Timperley's analysis of
feedback across more than 500 studies found that the most effective feedback is
specific, task-focused, and tied directly to the student's own work. Feedback at the
level of a general evaluation, such as "good structure but needs more analysis,"
produces far weaker learning effects than criterion-referenced, actionable comments
linked to particular sections of the work.
2
Under time pressure, it is precisely the
specific, criterion-referenced comments that lecturers are least able to write
consistently.
Weaver's study of student perceptions of written feedback found that students identify
four main characteristics of unhelpful feedback: comments that are too general or
vague, that lack guidance on how to improve, that focus on the negative without
explanation, or that are unrelated to the assessment criteria.
3
Each of these failure
modes is more likely to appear when the feedback is written quickly under deadline
pressure. Students notice the difference, and it is reflected directly in how they rate the
Assessment and Feedback category in the NSS.
"Students identify feedback as unhelpful when it is too general, lacks guidance, or is unrelated to the assessment criteria."
There is also a consistency problem that operates across a cohort rather than within a
single submission. Carless's survey across eight universities found that lecturers and
students hold systematically different perceptions of the feedback process: lecturers
believe they are providing more useful and detailed feedback than students experience
receiving.
4
Within a cohort, variation in feedback depth between submissions is
routine. When students compare notes and find that comparable pieces of work
received different quality of comment, it registers as unfair, and that perception feeds
directly into NSS responses on assessment fairness.
Why do institutions find the marking workload difficult to see
clearly?
Part of the answer is that workload models often undercount what marking actually
requires. A model that allocates 10 to 15 minutes per submission is, implicitly, a model
designed for brief comments rather than for the criterion-referenced feedback that
Black and Wiliam's foundational research shows produces meaningful learning gains.
5
When time is allocated for short feedback, lecturers who try to write useful feedback
exceed their allocation. When they adjust to fit the allocation, the feedback loses the
specificity that makes it useful.
There is also a visibility problem. When marks are submitted on time, the process
appears to have worked. The institution does not routinely see how long marking
actually took, what it displaced from the week, or what the quality of individual
feedback comments was. NSS data provides a lagging signal: students respond
months after the feedback experience, and even then the results are aggregated in
ways that obscure which modules or marking practices are driving the scores.
What approaches are institutions taking to address the marking
workload?
Several approaches are in use across UK higher education, with notably different
results in terms of both workload reduction and feedback quality.
Approach
What it involves
Evidence on effectiveness
Assessment
redesign
Reducing the number of
assessed pieces per module, or
shifting to formats requiring less
written feedback
Reduces volume but does not improve
the quality of feedback per submission.
May reduce learning opportunities
where formative assessment frequency
matters
Peer assessment
Students assess each other's
work against a rubric, with
academic oversight
Falchikov and Goldfinch's meta-
analysis of 48 studies found peer
marks diverge significantly from
lecturer marks when students assess
multiple criteria rather than making
holistic judgements
6
Structured
feedback
templates
Pre-designed comment banks or
templates that lecturers
complete rather than write from
scratch
Reduces writing time but can produce
generic-feeling feedback. Does not
resolve the consistency problem across
markers
AI-assisted
assessment
AI generates criterion-
referenced feedback for every
submission; the lecturer reviews
and approves before release
In Eduface's UK pilot programmes, AI
assessments aligned with lecturer
marks in 95 per cent of cases, with
students receiving specific, consistent
feedback across the full cohort
Of the approaches in use, AI-assisted assessment is the only one that addresses the
core problem directly: giving every student specific, detailed feedback without
requiring the lecturer to write it from scratch for each submission. The quality standard
is not reduced. It is maintained consistently across the cohort, which is precisely what
manual marking under time pressure fails to do.
What does AI-assisted assessment save in practice?
95%
In UK pilot programmes, Eduface's AI-generated
assessments aligned with lecturer marks in 95 per cent of
cases. Lecturers working through the review interface are
approving and refining feedback rather than writing it from
scratch, which is substantially faster and produces more
consistent output across the cohort.
Eduface UK pilot data, 2023–2024. Pilot partners include Bath Spa University
and De Haagse Hogeschool.
Lecturers using Eduface in pilot programmes report that the review phase takes a
fraction of the time that writing feedback from scratch required. The saving is not
uniform: submissions that diverge from the rubric, or that present arguments the
criteria did not anticipate, require more careful attention. But across a typical cohort,
the workload reduction is substantial. Crucially, the feedback students receive is
longer, more specific, and more consistently criterion referenced than what time
pressured manual marking produces. Nicol and Macfarlane-Dick identify criterion
referenced consistency as a prerequisite for students developing the capacity to
evaluate their own work: a skill that only develops when feedback is reliable and
structured enough to be used as a standard.
7
Eduface integrates with Blackboard, Brightspace, Moodle, and Canvas. Submissions
flow through the existing LMS, the AI generates structured feedback and provisional
marks, and the lecturer reviews everything in a single interface before releasing to
students. There is nothing new for students to learn and no change to the submission
process.
Where marking time goes: per submission
Without AI assistance
Reading (7 min)
Analysis (4 min)
Writing feedback (7 min)
Mark (2 min)
20 min
With Eduface AI assistance
Reading (7 min)
Analysis (4 min)
Review (3 min)
approx. 14 min
The writing phase is replaced by review. Reading and analysis remain with the lecturer, as they should.
AI assessment removes the repetitive writing phase, not the intellectual engagement with the work. Lecturers still
read, still analyse, still make the final call.
Frequently asked questions
Q
Will using AI assessment mean lecturers lose touch with student work?
Under time pressure, most lecturers already skim submissions rather than reading
them carefully. With AI-assisted assessment, the feedback is generated before the
lecturer opens each submission, so the task becomes reading the work and the AI
assessment together and deciding whether to approve or revise. Research by Carless
identifies feedback dialogue between lecturers and students as central to the
feedback process. AI assistance creates more time for that dialogue, not less.
Q
How much marking time does AI assistance actually save?
Lecturers in Eduface pilot programmes report reducing the marking phase from
roughly 20 minutes per submission to approximately 5 to 8 minutes for review and
approval. On a cohort of 80 students with two assessed pieces per semester, that is
the difference between 53 hours and around 13 to 21 hours of active marking work.
The reading and intellectual engagement with the work remains: the writing time is
what is reclaimed.
Q
Does AI feedback meet the quality standard that research recommends?
In Eduface's UK pilot programmes, AI assessments aligned with lecturer marks in 95
per cent of cases and generated criterion-referenced, task-specific comments of the
kind Hattie and Timperley identify as most effective for student learning. The feedback
students receive is longer and more specific than what time-pressured manual
marking consistently produces. The lecturer reviews every submission before release,
ensuring the standard is maintained across the cohort.
Q
Is the marking workload problem unique to the UK?
No. The arithmetic of cohort size multiplied by assessment frequency multiplied by
feedback time applies wherever higher education operates at scale. The problem is
particularly visible in the UK because NSS data makes the student experience of
feedback directly measurable. Dutch institutions piloting Eduface report the same
workload dynamics and the same improvement in feedback consistency once AI
assistance is introduced.
Q
Could reducing assessment volume solve the problem instead?
Reducing the number of assessed pieces per module does reduce marking volume,
but Black and Wiliam's research on formative assessment shows that frequency of
feedback is itself a driver of learning gains. Reducing assessment frequency to
manage workload trades one problem for another. AI-assisted assessment allows
both: more assessed work and better feedback on each piece, without increasing the
total time burden on lecturers.
The marking workload in higher education is not a problem that can be solved by
working harder or faster. The UCU's workload data and the research evidence on
feedback quality both point to the same conclusion: the current structure is
unsustainable, and its primary cost is borne by students in the form of feedback that is
too brief, too generic, or too inconsistent to support genuine improvement.
1,2,3
AI assisted assessment changes the structure rather than asking lecturers to absorb more within it.
See how much time you could reclaim
Create a free lecturer account and run Eduface alongside
your next batch of submissions. No contract and no time
limit.
Create free account
Request a demo
References
University and College Union. (2016).
Workload is an education issue: UCU workload survey report
2016.
UCU.
Hattie, J., & Timperley, H. (2007). The power of feedback. Review of Educational Research, 77(1), 81–
112.
Weaver, M. R. (2006). Do students value feedback? Student perceptions of tutors' written responses.
Assessment & Evaluation in Higher Education, 31(3), 379–394.
Carless, D. (2006). Differing perceptions in the feedback process. Studies in Higher Education, 31(2),
219–233.
Black, P., & Wiliam, D. (1998). Assessment and classroom learning.
Assessment in Education:
Principles, Policy & Practice, 5
(1), 7–74.
Falchikov, N., & Goldfinch, J. (2000). Student peer assessment in higher education: A meta-analysis
comparing peer and teacher marks. Review of Educational Research, 70(3), 287–322.
Nicol, D. J., & Macfarlane-Dick, D. (2006). Formative assessment and self-regulated learning: A model
and seven principles of good feedback practice. Studies in Higher Education, 31(2), 199–218.
Workload & Efficiency
How Much Time Do Lecturers
Spend Marking? The Numbers
Are Worse Than You Think
Most institutions have a rough sense that
marking takes a lot of time. Few have looked
carefully at exactly how much, or what that time
costs in terms of the feedback quality students
actually receive. The arithmetic is striking.
Eduface
·
9 min read
·
Lecturers & DVCs
It is late on a Sunday evening. There are 34
essays left in the queue. You have been marking
since Friday. The comments you are writing now
are noticeably shorter than the ones you wrote
on Friday: not because the work is simpler, but
because your capacity to write detailed, specific
feedback has a ceiling, and you passed it some
time ago. The UCU found that 26 per cent of
higher education staff already work more than 50
hours per week during term time.
1
The marking
workload is a structural problem with a direct
cost: the quality of feedback students receive.
How much time do lecturers spend marking?
A typical lecturer with two modules and 80
students per module will face 160 or more
assessed submissions per semester. At a
conservative 20 minutes per submission for
meaningful feedback, that is over 53 hours of marking per semester, on top of teaching,
research, and administration. The UCU found
that 26 per cent of higher education staff
already work more than 50 hours per week
during term time.
What does the marking workload look
like at module level?
The easiest way to understand the marking
burden is to calculate it at module level. The
numbers follow directly from standard cohort
sizes and assessment frequencies, without
additional assumptions.
Marking hours per semester: the arithmetic
COHORT SIZE
ASSESSMENTS
MIN PER SUBMISSION
TOTAL HOURS
40 students
× 2
× 20 min
27 hrs
80 students
× 2
× 20 min
53 hrs
120 students
× 3
× 20 min
120 hrs
Per module, per semester. Most lecturers teach two or more modules simultaneously.
These calculations assume 20 minutes per submission a
conservative figure for the kind of specific feedback research
identifies as most useful.
For a lecturer running two modules of 80
students each, the total exceeds 100 hours of
marking per semester before research, teaching
preparation, or administrative duties are counted.
The UCU found that 75 per cent of higher
education staff describe their job as stressful,
with unrealistic workload cited as the leading
cause.
1
What happens to feedback quality when
marking time runs out?
The response to sustained marking pressure is
almost always the same: comments get shorter.
Not because lecturers do not value feedback,
but because shorter comments are the only
variable that can be adjusted when time is fixed
and the submission queue is not shrinking.
Hattie and Timperley's analysis of over 500
studies found that the most effective feedback is
specific, task-focused, and tied directly to the
student's own work. Feedback at the level of
"good structure but needs more analysis"
produces far weaker learning effects than
criterion-referenced, actionable comments.
2
Weaver's study found that students identify four
characteristics of unhelpful feedback: too
general or vague, lacking guidance on how to
improve, focused on the negative without
explanation, or unrelated to the assessment
criteria.
3
Each is more likely under time pressure.
"Students identify feedback as unhelpful when it is too general, lacks guidance, or is unrelated to the assessment criteria."
Carless's survey found that lecturers and
students hold systematically different
perceptions: lecturers believe they are providing more useful feedback than students experience receiving.
4
Variation in feedback depth within a
depth within a Variation in feedback depth within a cohort is routine and when students compare notes, the inconsistency registers as unfair.
Why do institutions find the marking
workload difficult to see?
Workload models often undercount what marking actually requires. A model that allocates 10 to 15 minutes per submission is, implicitly, designed for brief comments rather than the criterion- referenced feedback that Black and Wiliam's research shows produces meaningful learning gains.
5
There is also a visibility problem. When marks
are submitted on time, the process appears to
have worked. The institution does not routinely
see how long marking actually took, or what the
quality of individual feedback comments was.
What approaches are institutions taking?
Several approaches are in use across UK higher
education, with notably different results in terms
of both workload reduction and feedback quality.
Approach
What it involves
Evidence on effectiveness
Assessment
redesign
Reducing the number of
assessed pieces per module
Reduces volume but does not
improve quality per submission
Peer assessment
Students assess each other's
work against a rubric
Falchikov and Goldfinch found peer
marks diverge significantly from
lecturer marks across multiple
criteria
6
Structured
templates
Pre-designed comment banks
lecturers complete
Reduces writing time but produces
generic-feeling feedback
AI-assisted
assessment
AI generates criterion
referenced feedback; lecturer approves before release
In Eduface UK pilots, AI aligned with
lecturer marks in 95% of cases, with
consistent specific feedback across
full cohort
Of the approaches in use, AI-assisted
assessment is the only one that addresses the
core problem directly: giving every student
specific, detailed feedback without requiring the
lecturer to write it from scratch for each
submission.
What does AI-assisted assessment save
in practice?
95%
In UK pilot programmes, Eduface's AI-
generated assessments aligned with lecturer
marks in 95 per cent of cases. Lecturers are
approving and refining feedback rather than
writing it from scratch.
Eduface UK pilot data, 2023–2024. Bath Spa University and De
Haagse Hogeschool.
Lecturers using Eduface report that the review
phase takes a fraction of the time that writing
feedback from scratch required. Across a typical cohort, the workload reduction is substantial and the feedback students receive is longer, more specific, and more consistently criterion- referenced.
7
Eduface integrates with Blackboard,
Brightspace, Moodle, and Canvas.
Submissions flow through the existing LMS. There is nothing new for students to learn.
Where marking time goes: per submission
Without AI assistance
Reading (7 min)
Analysis (4 min)
Writing feedback (7 min)
Mark (2 min)
20 min
With Eduface AI assistance
Reading (7 min)
Analysis (4 min)
Review (3 min)
approx. 14 min
The writing phase is replaced by review. Reading and analysis remain.
AI assessment removes the repetitive writing phase, not the
intellectual engagement.
Frequently asked questions
Q
Will using AI assessment mean lecturers
lose touch with student work?
Under time pressure, most lecturers already
skim submissions. With AI assistance,
feedback is generated before the lecturer
opens each submission, so the task becomes
reading the work and the AI assessment
together. AI assistance creates more time for
genuine engagement, not less.
Q
How much marking time does AI
assistance actually save?
Lecturers in Eduface pilots report reducing
the marking phase from roughly 20 minutes
per submission to approximately 5 to 8
minutes for review. On a cohort of 80
students with two pieces per semester, that is
the difference between 53 hours and around
13 to 21 hours.
Q
Does AI feedback meet the quality
standard?
In Eduface UK pilots, AI assessments aligned
with lecturer marks in 95% of cases and
generated criterion-referenced, task-specific
comments. Students receive longer and more
specific feedback than time-pressured
manual marking produces.
Q
Is the marking workload problem unique
to the UK?
No. The arithmetic of cohort size ×
assessment frequency × feedback time
applies wherever higher education operates
at scale. The problem is particularly visible in
the UK because NSS data makes the student
experience directly measurable.
Q
Could reducing assessment volume solve
the problem instead?
Reducing assessed pieces reduces marking
volume, but Black and Wiliam's research
shows that feedback frequency itself drives
learning gains. Reducing frequency trades
one problem for another. AI-assisted
assessment allows both: more assessed work
and better feedback on each.
The marking workload in higher education is not
a problem that can be solved by working harder
or faster. The UCU's workload data and the
research evidence on feedback quality both
point to the same conclusion: the current
structure is unsustainable, and its primary cost is
borne by students. AI-assisted assessment
changes the structure rather than asking
lecturers to absorb more within it.
See how much time you could
reclaim
Create a free lecturer account and run Eduface alongside your next batch of submissions. No contract and no time limit.
Create free account
Request a demo
References
University and College Union. (2016).
Workload is an
education issue.
Hattie, J., & Timperley, H. (2007). The power of
feedback. Review of Educational Research, 77(1), 81–112.
Weaver, M. R. (2006). Do students value feedback?
Assessment & Evaluation in Higher Education, 31(3),
379–394.
Carless, D. (2006). Differing perceptions in the feedback
process. Studies in Higher Education, 31(2), 219–233.
Black, P., & Wiliam, D. (1998). Assessment and classroom
learning. Assessment in Education, 5(1), 7–74.
Falchikov, N., & Goldfinch, J. (2000). Student peer
assessment. Review of Educational Research, 70(3),
287–322.
Nicol, D. J., & Macfarlane-Dick, D. (2006). Formative
assessment and self-regulated learning.
Studies in
Higher Education, 31
(2), 199–218.