How to Handle Supervisor Feedback on Your Thesis
A chapter comes back with forty comments, half of them three words long. The problem is rarely that the feedback is wrong — it is that it arrives compressed, and you have to decompress it before you can act. Here is a method for doing that without rewriting everything.
Every comment is answering one of three questions
Hattie and Timperley's review of feedback research (Review of Educational Research, 2007) frames effective feedback as an answer to three questions: Where am I going? (the goal), How am I going? (the gap between the draft and the goal), and Where to next? (what to do about it). They call these feed up, feed back and feed forward.
Supervisor comments almost always answer the middle question and skip the other two. "Unclear" is a pure gap statement: it tells you something is wrong and nothing about the standard being missed or the move that would fix it. So before you edit, label each comment with the question it answers — and treat every comment that only says "how am I going" as an unfinished instruction you are entitled to complete by asking.
Then check what level it is aimed at
The same review distinguishes four levels feedback can target: task, process, self-regulation and self. The level tells you how much of the chapter is actually in question:
- Task — "This citation is from 1998; is there anything newer?" A local fix. Do it and move on.
- Process — "You are summarising each study in turn." Not a local fix. The method that produced the section is what needs changing, and every similar section has the same flaw.
- Self-regulation — "What did you decide about the sampling issue we discussed?" They are handing the judgement back to you on purpose. Answer it; do not wait to be told.
- Self — "Good effort." Hattie and Timperley found this the least effective level, because praise aimed at the person carries almost no information about the work. Enjoy it, then look for the comment underneath it.
The single most expensive mistake is treating a process-level comment as a task-level one — patching the three sentences that were marked, when the supervisor was describing a pattern that runs through thirty pages.
Sort before you edit: three buckets
Open the marked-up file, read every comment once without changing anything, and put each into one of three buckets in a separate document:
- Structural — the argument, the order of sections, what the chapter is for. Costly, and everything else depends on it.
- Evidential — missing sources, a claim without support, a reference that does not say what you claimed, an inconsistent reference list.
- Surface — wording, tense, formatting, typos.
Work strictly in that order. Polishing a paragraph you are about to delete is the most common way a week disappears, and structural changes routinely dissolve a dozen surface comments on their own.
A worked example
The comment:
"This reads like a list. Where is your voice?"
Decoded: a process-level gap statement, no feed forward. The draft:
Ahmed (2019) found a positive association between X and Y. Lin (2021) also found a positive association. Okoro (2022) reported no association.
The revision groups by claim and states the consequence:
Two cohort studies report a positive association between X and Y (Ahmed, 2019; Lin, 2021), while the single trial to test it directly found none (Okoro, 2022). The discrepancy tracks study design rather than population, which is why this thesis treats the observational evidence as suggestive rather than confirmatory.
Same three sources, same word count, and the paragraph now does something. If you are unsure whether your own prose has this problem, our guide to writing a literature review covers the grouping method in full.
"Your citations are inconsistent" is a fifteen-minute problem, not a rewrite. Run the entries through our free citation generator in one style — APA 7, MLA 9, Chicago, Harvard or IEEE — and the whole bucket clears at once.
Turn the buckets into a change log
Do not arrive at the next meeting with a clean draft and no account of it. Bring a two-column table: the comment on the left, what you did on the right, one row per comment. Three things happen. Your supervisor can see their feedback was used without rereading the chapter. Disagreements become explicit — "kept this, here is why" is a legitimate row, and it gets discussed instead of resurfacing at the viva. And by the time you write up, you have a record of every decision you made and the reason for it.
The format is deliberately the same one journals expect later; if you want the fuller version, see how to respond to reviewer comments.
The part nobody trains you for
Carless and Boud (2018) describe feedback literacy as four connected capacities: appreciating feedback, making judgements about quality, managing affect, and taking action. The third one is why a page of tracked changes can be genuinely hard to open — and why the standard advice to "not take it personally" is useless as an instruction. What works is procedural: read it once, close the file, come back the next day and reread it looking only for verbs. What am I being asked to do? A comment that survives that reading is information. A comment that does not is usually a tone problem, not a task.
One caveat worth holding onto: silence is not approval. If a chapter comes back with no comments on the argument, that may mean the argument is fine — or that nobody read it that closely. Ask directly.
Fewer comments to handle in the first place
CiteGood drafts from real papers and checks every claim against the source it cites — so the evidential bucket is close to empty before your supervisor opens the file.
Try CiteGood free →Frequently asked questions
What does it mean when a supervisor says a section is 'too descriptive'?
It means you have reported what the sources say without doing anything with it. Descriptive writing lists studies one per paragraph; analytical writing groups them by claim, states where they agree and disagree, and says what follows for your own question. The fix is usually structural — regroup the paragraphs around arguments rather than around authors — not a matter of adding stronger adjectives.
How do I ask my supervisor for clearer feedback without sounding defensive?
Ask a specific, answerable question instead of asking them to explain the comment. 'You marked this section as thin — would you want two more empirical studies here, or a sharper link back to the research question?' gives them a choice to react to. Sending the question a day or two before the meeting also gives them time to reread the passage.
What if my two supervisors give me contradictory feedback?
Do not silently pick one — that is how you end up defending a choice neither of them made. Write both comments side by side in a short note, state which one you propose to follow and why, and send it to both before you revise. Contradictions are usually a disagreement about the scope of the thesis, and surfacing it once saves you rewriting the same chapter twice.
Do I have to accept every change my supervisor suggests?
No. It is your thesis and your defence, and a supervisor who suggests a change is usually flagging a problem rather than prescribing the only cure. You can decline a specific suggestion if you can say what the underlying problem was and how you solved it differently. What you cannot do is quietly ignore the comment and hope it does not come back at the viva — it will.
How long should I wait before responding to harsh feedback?
Long enough to stop reading it as a verdict on you, which for most people is a day. Then reread it looking only for instructions. Carless and Boud (2018) treat managing that emotional reaction as a distinct skill within feedback literacy, alongside judging quality and acting on it — so the delay is part of the process, not procrastination.