Flattery feels nice. It also wastes your prep time.
When a tool tells you "Band 8, well done!", you learn one thing: nothing actionable. You don't know which criterion is strong, which is dragging you down, or what to do next. You close the tab feeling good and walk into the exam making the exact same mistakes. Inflated praise isn't kindness — it's a score you can't act on, sold to you as encouragement.
Real feedback should make you a little uncomfortable and a lot clearer. The test isn't whether it flatters you; it's whether you finish reading it knowing precisely what to fix.
The three things real feedback must contain
1. A band for each of the four criteria. IELTS Writing is marked on Task Response, Coherence & Cohesion, Lexical Resource, and Grammatical Range & Accuracy — four separate scores, equally weighted. A single overall number hides which one is costing you. "Band 6.5" tells you nothing; "Task Response 6, Grammar 7" tells you where to spend the next hour.
2. The exact evidence — your own words, quoted back. A claim like "your vocabulary is limited" is an opinion. "You wrote 'this is a very big problem for many people' three times" is evidence. Feedback that can't point at the sentence it's judging is just an impression, and impressions can't be trusted or learned from.
3. The single smallest change that lifts the band. Not ten things — one. The highest-leverage fix, named plainly: "Add a clear position in your introduction and your Task Response moves from 6 to 7." One concrete next move beats a wall of generic advice you'll never apply.
Vague vs useful: the same essay, two responses
Vague: "Good effort! Your essay is well-structured and uses good vocabulary. Try to add more detail. Estimated band: 7.5." Pleasant. Useless. You can't do anything with it.
Useful: "Task Response 6 — the prompt asks whether the benefits outweigh the drawbacks, but you list both sides without ever stating which wins; you wrote 'there are advantages and disadvantages' and stopped. Coherence 7. Lexical Resource 7. Grammar 6 — recurring article slips, e.g. 'government should invest in education'. Smallest fix: state a clear position in sentence one of your conclusion. That alone likely moves Task Response to 7."
One of these tells you exactly what to rewrite tonight. The other tells you you're lovely.
Why Task Response is the part most tools skip
It's the hardest thing to judge automatically, so most free checkers quietly skip it. Counting commas and flagging spelling is easy; deciding whether you actually answered the whole question — took a clear position, addressed every part of the prompt, developed relevant ideas — requires reading for meaning, not pattern-matching surface features.
That's exactly why it matters most. A grammatically spotless essay that doesn't answer the question is capped hard on Task Response, and since the four criteria are weighted equally, that one ceiling drags your whole band down. If your feedback never mentions whether you answered the question, it's ignoring the criterion most likely to be holding you at Band 6.5.
Re-marking is where the score actually moves
One mark is a snapshot. Improvement comes from the loop: read the feedback, make the single named change, re-submit, and watch which band moves. If Task Response was 6 and your fix pushes it to 7, you've just proven the diagnosis was right — and learned a rule you'll carry into the exam. If it doesn't move, you misread the fix, which is also useful to know.
This is why honest, criterion-level feedback beats a flattering one-off score every time. You can only iterate against marking you can trust. Inflated numbers break the loop, because there's nothing real to push against.
How to take honest feedback without being demoralised
A band is directional, not a verdict on you. "Task Response 6" doesn't mean you're a 6 of a person — it means this essay, today, didn't state a clear position, and that's a fixable, ordinary mistake. Separate the score from your self-worth and it becomes what it actually is: a map.
The candidates who improve fastest treat low marks as information, not insults. They read the evidence, make the one fix, re-submit, and move on. Honest feedback can sting for a second; vague flattery costs you the whole exam.