What the feedback actually means
Description tells the reader what happened, what a source says, or what a theory claims. Analysis tells them why it matters to the question you were asked.
Most essays marked "descriptive" are not badly written. They are usually well organised, properly referenced, and demonstrably the product of a lot of reading. That is exactly what makes the feedback so frustrating — the work went in, and the marks did not come out.
The gap is this. A marker reading a descriptive paragraph learns that you did the reading. A marker reading an analytical paragraph learns that you understood what the reading was for. Only the second moves you up a band, because UK band descriptors almost always separate on judgement rather than coverage. A 2:2 and a first can contain the same facts.
If a sentence would still be true whatever the essay question was, it is probably description.
That test is worth applying literally. Take any sentence from your draft and imagine it appearing in an essay on a related but different question. If it would sit there unchanged and unremarkable, it is doing no work for the question in front of you.
Why it happens to people who read a lot
Description is not laziness. It is usually the opposite: the more you read, the more you have to report, and reporting feels like the responsible thing to do. There is also a structural cause. Most of the sources you read are themselves descriptive at the point you encounter them — a textbook chapter, a case summary, a methods section — so the register you absorb is the register you reproduce.
Three habits push in the same direction:
- Fear of being wrong. Description is safe. It is hard to be marked down for accurately stating what a source says, so it feels like the low-risk option. In practice it is the high-risk option, because it caps your grade.
- Writing in the order you read. If your paragraphs follow the sequence of your reading list, the structure is a record of your research rather than an argument.
- Treating the word count as something to fill. Description expands effortlessly. Analysis does not, so when a deadline is close, description is what arrives.
The same evidence, twice
These two paragraphs use identical source material, for the question To what extent did wartime rationing change British diets?
Rationing was introduced in January 1940 and initially covered bacon, butter and sugar. It was later extended to meat, tea, jam and cheese. The Ministry of Food ran extensive campaigns encouraging households to grow their own vegetables. Bread was not rationed until 1946, after the war had ended.
Why. Accurate, well ordered, correctly sequenced — and it answers nothing. Every sentence is a fact in a queue. Nothing here addresses to what extent, and the paragraph would sit unchanged in an essay on the administration of the Ministry of Food.
Rationing's dietary impact was uneven by design. The staples restricted from 1940 were concentrated in fats and sugars, while bread remained unrationed until 1946 — suggesting the Ministry was protecting caloric floors even as it deliberately reshaped diet composition. The change was therefore real, but narrower than the campaigns implied.
Why. Same facts, same length. The dates now carry an argument: the contrast between what was restricted and what was not becomes evidence for a claim about intent. The paragraph lands on the judgement the question asked for.
Notice what did not change. No new reading was needed. No longer word count. The analytical version is not more sophisticated in vocabulary — it is arranged around a claim rather than a chronology.
Three habits that shift the balance
Make the last sentence do work
If a paragraph ends on its final piece of evidence, it ends descriptively. Write one more sentence saying what that evidence establishes for your argument. This single habit changes more essays than any other, because it forces the interpretive step you were probably making silently in your head.
Weak endings look like: "…and this was repeated in the 1994 study." Strong endings look like: "…which suggests the effect is not an artefact of the original sample."
Put the comparison inside the paragraph
Description handles one thing at a time. Analysis tends to emerge when two pieces of evidence sit in the same paragraph and you have to explain the tension between them. If your draft has one source per paragraph throughout, that is a structural signal — you are summarising a reading list rather than building a case.
Use the question's own verb
If the question says evaluate, some of your sentences must actually evaluate. Echoing the question's verb is a crude but reliable way to check you have done the task set rather than an adjacent one you found easier. It is also what the marker is scanning for.
A caution. Analysis is not the same as opinion. "This policy was clearly misguided" is not analysis unless the evidence and reasoning that make it clear are on the page. Unsupported assertion is marked down at least as hard as description — and often harder, because it looks like overreach rather than caution.
Where description is correct
Some sections are meant to be descriptive, and stripping description out of them is its own error. Methods sections describe. Case facts in a law problem question describe. A literature review describes what studies found before it analyses what the pattern means. Background paragraphs establish shared ground so the analysis has something to stand on.
The problem is not description existing. It is description arriving where judgement was expected, and the ratio being wrong across the piece as a whole. As a rough guide: if you cannot point to your argument in a given paragraph, that paragraph needs to earn its place some other way.
Paste a paragraph of your own. This measures how your writing behaves — sentence variety, hedging, nominalisation and the phrases that read as filler.
Results appear here. Around 60 words gives the most reliable reading — shorter passages make sentence-variety scores jumpy.
A checklist before you submit
- Every paragraph ends on a claim, not a fact
- The question's verb appears in your own sentences, doing real work
- No paragraph presents evidence without saying what it establishes
- At least one paragraph holds two sources in tension
- Nothing in the essay would survive unchanged under a different question
- Every judgement has visible evidence behind it, not just confidence
If a draft fails four of six, the fix is usually structural rather than sentence-level — the essay is arranged around topics instead of around an argument, and rewriting individual sentences will not move it far.