Most marks are not lost in one dramatic collapse. They leak away quietly, in the same handful of places: a thesis that never quite commits, a quotation left to speak for itself, a paragraph that retells the reading and then stops. Markers see these habits constantly, and each one has a fix you can apply in a single editing pass. Here are the five that cost students the most, with something practical to do about each.

1. The thesis that refuses to commit

Compare these two openings:

• Weak: "This essay will explore the role of social media in political debate."
• Strong: "Social media has not widened political debate so much as rewarded the loudest voices within it."

The first names a topic. The second makes a claim that someone could reasonably argue with — and that is the point. A marker can only award analysis marks if there is an argument to analyse. Words like "explore" and "discuss" describe what you are about to do rather than what you think, so they announce an essay that is still deciding.

The fix takes two minutes. Draft your thesis as it comes, then rewrite it as a sentence beginning "This essay argues that…" — even if you delete those four words later. Then test it: could a sensible person disagree? If not, you have written a summary, not a position. Finally, check that your conclusion makes the same claim. A thesis that quietly changes shape between the introduction and the conclusion reads as confusion, not nuance.

2. Evidence that is quoted but never used

This is the most common single fault in undergraduate writing. A source appears, does its turn, and the essay moves on.

As one scholar writes, "the reform was widely unpopular."

Nothing has been done with that sentence. Unpopular with whom? Measured how? Does "widely" hold across regions, classes, or the whole period under discussion? Evidence does not carry meaning by itself; you have to show the reader what you see in it.

After every piece of evidence, write at least two sentences of your own: one that unpicks a specific word or detail, and one that links the point back to your claim, or to the counter-argument you are answering. A workable ratio is one sentence of evidence to two or three of analysis. Where you have a choice, prefer precise evidence over impressive evidence — a short quotation you can dissect properly beats a long one you can only gesture at.

3. Retelling instead of arguing

Some essays read as a competent summary of the reading, with the argument left implied. The tell is a paragraph that could sit in an encyclopaedia entry: dates, events, descriptions, no judgement. It is accurate, and it earns very little.

Apply the "so what?" test to every paragraph. If the honest answer is "it provides background", either compress it into a single sentence or find the point it is proving. Signposting phrases help you build the habit: "This matters because…", "The significance of this is…", "What this account obscures is…". They are scaffolding, not style, so use them freely while drafting and trim the clumsiest ones at the end.

Chronological structure deserves a warning too. "And then… and then…" feels safe because it is easy to follow, but it rarely builds an argument. Order your points by strength of reasoning, not by the order in which you read them.

4. Paragraphs that wander

A paragraph should make one move in the argument. If it makes three, the reader loses the thread. If it makes none, they wonder why it is there.

Four quick tests:

• Read only your first sentences, in order. Do they form a coherent argument on their own? If not, your structure is the problem, not your prose.
• Look for paragraphs running past about two-thirds of a page. They usually contain two ideas competing for the same space.
• Check that each paragraph ends somewhere different from where it started — a small advance, not a restatement.
• Make sure linking sentences refer to the actual previous point rather than a vague "another important factor is…".

If a paragraph fails these tests, write its topic sentence at the top of a blank page and move the remaining sentences under whichever heading they genuinely belong to. Rearranging is nearly always faster than rewriting, and it exposes gaps in the argument before a marker finds them.

5. Small slips that cost expression marks

These rarely ruin an argument on their own, but they add up. Tense drifting between past and present. "Etc." closing a list you could not finish. Contractions and conversational phrasing in formal prose. Second person ("you") in an academic essay. Titles without italics. References formatted three different ways across four pages.

Hedging is the costliest of the group. "Arguably", "somewhat" and "it could be said that" drain confidence from a claim you have already earned through evidence. Say the thing, then support it.

The fix is a personal error list. When an essay comes back, copy every recurring comment into one document. Before you submit anything, read that list and check the draft against it. Read the final version aloud as well; you will catch tangled sentences and dropped words that silent reading hides completely. Tidy referencing as you go rather than at midnight, when you are least likely to notice that a citation has lost its page number.

A ten-minute check before you submit

Work through this in order, with the essay in front of you:

1. Read the introduction and the conclusion back to back. Do they make the same claim?
2. Highlight every quotation or piece of data. Does each one have analysis after it?
3. Read only your topic sentences. Is there a line of argument running through them?
4. Search for "very", "really", "a lot", "things" and "society". Replace what you can with something specific.
5. Check the reference list against your in-text citations, one by one.
6. Read the first paragraph aloud. If you stumble, your reader will too.

None of this is about writing more. It is about giving the marker fewer reasons to hesitate, which is what a surprising share of the available marks is actually for. Fix the thesis, use the evidence, argue rather than describe, keep each paragraph to one job, and read the thing once more before you hand it in.

Photo: andreapassaro / Pixabay