What the introduction settles is scope, not impression
One of the comments that comes back most often on a returned assignment is that it drifted off topic. You find the stray paragraph, delete it, and the same comment shows up on the next assignment. The cause was never in that paragraph. When the introduction fixes no ground to stand on, the body just puts out whatever the writer knows that looks usable, in no particular order.
So an introduction is less a greeting for the reader than a written promise about what you will and will not do. The narrower the promise, the easier the body gets. Say you will cover the impact of artificial intelligence on society and any paragraph fits while any cut feels like a loss. Say you will compare three university libraries that added a chatbot, using one shared set of criteria, and what belongs and what goes is decided on the spot.
People disagree about whether the introduction comes first or last, and the answer splits in two. The three lines that fix the scope come first. Turning those three lines into finished prose comes after the body is done, because reading the sources shifts the scope a little as you go.
The three moves came out of taking research introductions apart
The linguist John Swales gathered research article introductions from across disciplines in his 1990 book Genre Analysis and pulled out the moves that kept repeating. What came out of that is the three-step pattern known as the CARS model, short for creating a research space. It was drawn from journal articles, but it transfers to coursework intact, because what a reader wants to confirm in an unfamiliar piece of writing does not change with length.
- Establish a territory. Say which conversation this topic sits in, summarising what is already known or already studied in a couple of sentences.
- Mark the gap. Point at what is still unsettled inside that conversation. Something earlier accounts missed, results that contradict each other, or a question nobody has asked will each do.
- Occupy the gap. State what this piece does about it: what you compare, which sources you use, and the order the sections will run in.
The second move is the one that goes missing
Summarise what you learned, jump straight to the conclusion, and the reader follows the body without knowing why the piece exists. The sentence that marks the gap does not need to be long. Something like: the two accounts above each cover one side, and material that puts them side by side was hard to find. That is enough.
The two introductions below were written for the same assignment. The first stops at move one; in the second, each line carries one move.
Body structure splits two ways, inquiry or argument
For an assignment that reports something you investigated or measured yourself, use introduction, methods, results, discussion. It looks like an old convention, but it settled in fairly recently. A survey that sampled articles from four medical journals between 1935 and 1985 found the four-part structure first appearing in the 1940s, reaching 80 percent in the 1970s, and becoming effectively the only pattern in original papers through the 1980s (Sollaci and Pereira, Journal of the Medical Library Association, 2004).
For an assignment where you read and build your own case, the four parts give way to claim, evidence, counter-argument, response. Mix the two and you get a methods section carrying a claim and a discussion section that ends in a summary. The wording of the prompt tells you which one you are in. Investigate, measure, compare points to the first; discuss, evaluate, give your view points to the second.
| Wording of the prompt | Structure to use | What the closing section carries |
|---|---|---|
| Investigate, measure, report the results | Introduction, methods, results, discussion | How the results answer the opening question, and what this method could not show |
| Discuss, evaluate, give your view | Claim, evidence, counter-argument, response | The claim that survives the counter-argument, and the conditions it holds under |
| Compare, sort into types | Criteria, case by case, a comparison table | The differences the criteria separated out, and what those criteria hide |
A citation is an address for evidence, not a courtesy
Spend enough time matching a citation format and citing starts to feel like rule-following. Citing does one thing: it leaves an address so the reader can go and check the sentence themselves. A citation nobody can follow is worth nothing no matter how correct the format is.
The easiest place to lose that value is a second-hand citation. Take a statistic a blog quoted and cite the blog for it, and the reader has no way to learn where the number came from. Use blogs and encyclopedia entries as a starting point, then put the original they link to in your footnote. If you cannot find the original, dropping the number is the better move. One unverified figure pulls down the credibility of every sentence around it.
- Quoted sentences go inside quotation marks, paraphrases go without. Both still carry a source.
- Note the date you checked next to figures and dates. Web sources change without telling anyone.
- If the syllabus names a citation style, that one wins. If it does not, staying consistent within the one piece is enough.
Length is something you divide, not something you fill
Given a five-page requirement, most people start at page one and write forward. The introduction swells to a page and a half and the conclusion becomes three lines written thirty minutes before the deadline. Dividing first makes that less likely. For five pages, half a page of introduction, three pages of body, half a page of conclusion and a page of references sits about right. Split those three body pages across three sections and each section is one page, three or four paragraphs.
Divided this way, the thin spots become visible. If the second section stalls at half a page, that is not a cue to pad sentences, it means the claim in that section rests on a single piece of evidence. Find one more source and both the length and the persuasion go up together. Half a page filled by stretching sentences is the part a reader spots first.
The last thirty minutes before you submit
Right after finishing, your own writing reads well, because the information that bridges the sentences is still in your head. The fastest way to sit in the seat of someone without that information is to read only the headings, top to bottom. If the argument flows through the headings alone, the body usually flows too.
- Whether the three lines of the introduction each carry one move
- Whether every body heading pairs with one promise made in the introduction
- Whether you actually opened every source you cited
- Whether the conclusion answers the opening question instead of summarising
- Whether the file name, format and submission channel match what the syllabus asked for
What to try first on the next assignment
On the next assignment, write the three lines of the introduction before you collect anything. They do not need polish: one line of territory, one of gap, one of what you will do. Take those three lines to your instructor or teaching assistant and ask whether that scope matches what the assignment is after, and you will less often have to redraw the scope after reading everything. Writing the three lines takes about ten minutes.
Frequently asked questions
Is it okay to draft with a generative AI tool?
It depends on the school and the course. Even inside one school, some courses allow it as long as you note which tool you used, and others ban it outright. So the answer lives in the syllabus and the assignment notice. If neither says anything, ask whoever runs the course and keep the reply in your mail. Even where it is allowed, one more check remains: whether the facts and citations in the draft actually exist. Paper titles that were never published and page numbers that lead nowhere come out in convincing formatting more often than you would expect. Open each cited source yourself, and drop anything you cannot find.
How many references does a report need?
The only place a raw count is the standard is a course whose syllabus names a minimum, and there that number is a floor, not a target. Everywhere else, placement matters more than count. Check whether every point in the body that produces a fact or a figure carries an address to verify it. Ten items in the list against two citations in the body means the other eight are a badge saying you read them, not evidence doing work. The reverse case, many citations drawn from two or three sources, produces a piece that carries one viewpoint across.