A Research Paper Outline You Can Fill In Today
- A research paper has seven sections: introduction, literature review, method, results, discussion, limitations, conclusion.
- Each section answers one question. Introduction asks why this matters. Results says what happened. Discussion says what it means.
- Keep results and discussion apart. Reporting a number and arguing about it are two different jobs, and mixing them is the most common error.
- A literature review is an argument about the field, not a stack of summaries. It has to end at the gap your question fills.
- AI can shape an outline and explain a concept. It cannot be trusted with citations, because invented references are common and they look real.
You have a deadline, a rough idea, and a blank document. The hard part is not the writing. It is knowing what belongs in each section, so you stop rewriting the same paragraph in three places.
This page gives you the standard structure, a plain explanation of what each part is for, and a template you can paste into your document right now.
The seven sections, in plain words
Introduction and literature review: the setup
The introduction exists to make the reader care. It names a real problem, says why it matters now, and ends with your question. Not a topic. A question. "Sleep and grades" is a topic. "Does bedtime consistency predict first year grades?" is a question you can answer.
The literature review exists to show a gap. You read what others found, and you report where they agree, where they clash, and what nobody has checked. The gap is the point. If your review does not end with something missing, your paper has no reason to exist.
Method and results: the record
The method exists so a stranger could repeat your study. Who took part, how many, what you gave them, in what order, how you measured, how you analysed. A good test: could someone in another country run this from your description alone? If not, add the missing detail.
The results section exists to report what you observed. Numbers, tables, figures. Nothing else. You are a camera here, not a lawyer. Write "scores rose by 12 points", not "scores rose, which shows the method works".
Discussion, limitations and conclusion: the meaning
The discussion is where you finally argue. Answer your question directly in the first line. Then tie your findings back to the studies in your review. Say who you agree with and who you contradict. Explain anything surprising instead of hiding it.
Limitations is where you say what you could not do. Small sample. One school. Data people reported about themselves. Six weeks, not a year. Naming these makes you more credible, not less. Markers notice when a paper pretends to have no weak spots.
The conclusion answers one question: what changes now? Not a summary of what you already wrote. Say what someone should do differently, or what the next study should test. One tight paragraph beats a page.
What each section does, and how it goes wrong
| Section | Its job | Common mistake | Share of length |
|---|---|---|---|
| Introduction | Show the problem and ask the question | Explaining background for pages before asking anything | 10% |
| Literature review | Map the field and find the gap | A list of summaries with no argument | 20% |
| Method | Let someone repeat the study | Missing the boring detail that makes it repeatable | 15% |
| Results | Report what you observed | Interpreting the numbers instead of reporting them | 20% |
| Discussion | Say what the findings mean | Repeating the results table in sentences | 25% |
| Limitations | Name what you could not do | Skipping it, or listing only harmless flaws | 5% |
| Conclusion | Say what changes now | Rewriting the introduction in past tense | 5% |
Example allocation, to show the shape of a balanced paper. These are working guides, not a rule. Your department's brief always wins, and lab reports often shrink the review and grow the method.
Copy this template into your document
Paste this, then fill each block. The line under every heading tells you what goes there. Delete the prompt lines as you replace them.
Working title:
Research question:
One sentence. A question, not a topic.
1. INTRODUCTION (about 10%)
The problem, why it matters now, your question, and one line on
what this paper does. End on the question in plain words.
2. LITERATURE REVIEW (about 20%)
Group studies by what they claim, not by author. Say where they
agree, where they clash, and what nobody has tested. Finish on
the gap your question fills.
3. METHOD (about 15%)
What you did, in order. Who took part and how many. Materials,
steps, how you measured, how you analysed. Enough that a
stranger could run it again.
4. RESULTS (about 20%)
What you observed. Numbers, tables, figures. The finding that
answers your question goes first. No "this suggests" here.
5. DISCUSSION (about 25%)
Answer your question in the first sentence. Compare with the
studies from section 2. Explain the surprises.
6. LIMITATIONS (about 5%)
What you could not do, and what that does to your claim.
7. CONCLUSION (about 5%)
What changes now that this paper exists. One paragraph. No new
evidence, no new citations.
REFERENCES
Every source opened and checked by you. Authors, year, title,
and a link or DOI that resolves.
If you want a first draft of the headings for your own question, the AI Outline Generator will produce a section list you can edit. Treat what it gives you as a starting shape, then rewrite it against the template above.
Three mistakes that cost the most marks
Results that argue
This one is everywhere. A student writes "the group improved by 12 points, showing that spaced practice works." Two jobs in one sentence. Split it. The 12 points belong in results. "Showing that spaced practice works" belongs in discussion, with the studies that back it up.
Quick test: read your results section and highlight every word like suggests, proves, because, importantly. Each one is a piece of discussion that wandered into the wrong room. Move it.
A literature review that is just a list
The weak version reads like this. Smith found X. Jones found Y. Patel found Z. Every paragraph starts with a name and ends with a summary. Nothing connects.
The strong version groups by claim. "Three studies link bedtime to grades, but all three used students who volunteered. Two studies that used whole cohorts found no effect." Now the reader sees a disagreement, and the sampling method is the suspect. That is an argument, and it points at a gap.
Rule of thumb: if you could reorder your review paragraphs at random and nothing broke, it is a list. A real review falls apart when you shuffle it.
A conclusion that repeats the introduction
Your introduction says nobody knows the answer. Your conclusion says what the answer is. If both paragraphs could swap places without confusing anyone, you wrote the same thing twice.
Try this. Cover your introduction and write the conclusion from one prompt only: because of this paper, someone should now do X. That forces something new onto the page.
Where AI helps here, and where it lies
Be honest about this, because your department will be. AI is genuinely good at three jobs on a research paper. It can propose a section structure for your question. It can explain a concept or a statistical test you half remember. It can tighten a paragraph you have already written from your own work.
There is one thing it must never do. It must not supply your citations. Language models produce references that look completely real, with plausible authors, a believable journal, a well formed DOI, and no existing paper behind any of it. This failure is common, it is well documented, and graders now check for it. A fabricated reference is treated as academic misconduct, not a typo.
So the rule is simple. Every source in your reference list is one you opened yourself. If you cannot find the paper in a library database or on the publisher's site, it does not go in. Our guide How to Use AI for Research walks through the checking habit in more detail.
What does work well: bring the real paper to the model. Once you have a PDF you found yourself, you can ask for a plain summary of its method, or ask what its sample size limits. The guide How to Summarize PDFs with AI covers how to do that without losing the detail you actually need.
Run this before you submit
Fill in one block tonight
Paste the template into your document now. Then write only two things: your research question as a single sentence, and the gap line at the end of the literature review. Those two decide the shape of everything else.
If your question still feels like a topic, say it out loud to a free AI Chat and ask it to turn your topic into three testable questions. You get one question with no signup, and a free account adds 100,000 starter tokens, which is plenty for shaping questions but thin once you start pasting whole sections and papers, so the paid plans fit a full write up better. Pick the question you can actually answer with the data you can get, then fill the next block tomorrow.