The conclusion is not a place to finish — it is a place to land. Weak conclusions fade out, summarize the paper the reader has just read, or apologize for what they did not do. Strong conclusions do something different: they synthesize the argument, state what changes because of it, acknowledge limits honestly, and point forward to the next question. This guide shows you how to write one — the five moves a research paper conclusion makes, the difference between synthesis and summary, and the sentences that should never appear in a final paragraph. You will see a worked example, learn the failure modes that cost marks, and pick up a drafting pattern you can revise into the conclusion your paper actually deserves.
PaperDraft is a writing assistant, not a paper generator — the draft is your starting point, not your submission. You are responsible for editing, verifying sources, and following your school's academic integrity policy.
What a research paper conclusion is — and what it is not
A research paper conclusion is the closing section that does five things in compact form: it restates the thesis in new language, synthesizes what the paper has shown, states implications, acknowledges limitations, and points to future work. It is typically one to two paragraphs in a short paper, up to a full section in a longer one. It is the place where the reader finishes with a clear sense of what the paper argued, why it matters, and what remains open.
A conclusion is not a summary. A summary walks section by section through the paper reminding the reader what they just read. That is boring and unnecessary — the reader remembers. A conclusion synthesizes: it combines what the sections have shown into a single claim about the paper as a whole.
A conclusion is not an apology. "Due to the limited scope of this paper, I was unable to address X, Y, and Z" reads as weak. Limits are real and should be named, but name them as honest scope decisions, not as failures.
A conclusion is not a place for new evidence. If a citation, a dataset, or an argument is important enough to include, it belongs earlier in the paper where it can be developed. Introducing new evidence in the conclusion leaves it undefended and signals that the paper's structure was never quite finished.
A conclusion is not a repeat of the introduction. The introduction argues for why the question matters; the conclusion argues for what the paper has established about the answer. They are different moves. A conclusion that reads like a paraphrase of the introduction wastes the reader's attention at the moment it is most valuable.
Finally, a conclusion is not a throwaway paragraph written at 2 a.m. Readers — including graders — often read the introduction and conclusion first to decide what the paper is. Treat it as a load-bearing section, not a closing formality.
Before you write
Draft the conclusion last, after the body of the paper is stable. A conclusion written from a plan rather than from a finished draft will describe the paper you intended, not the paper you produced — and the two are usually different.
Before opening the document, gather:
- The actual thesis as supported by your finished paper. This may differ from the thesis you wrote in the introduction. If so, revise both to match. Your thesis statement guide covers how to sharpen it.
- The two or three strongest findings from the body. These are what the synthesis combines.
- Honest limits. Sample size, scope, method, time window, evidence type. List them for yourself before deciding which belong in the conclusion.
- One or two specific next questions the paper opens. Not "more research is needed" — an actual question.
Check the conventions of your genre. Empirical papers often have a combined Discussion/Conclusion that interprets results in depth, with a short concluding paragraph that lands the paper. Humanities papers often have a longer, more literary closing section that revisits the opening framing. Lab reports may have a minimal conclusion restating hypotheses, results, and implications in two or three sentences. Read recent examples in your target venue or from your instructor's feedback on prior papers.
Know the expected length. In a 3,000-word paper, a conclusion of 150–300 words is typical. In a 6,000-word paper, 300–600 words. In a thesis chapter, a full section. Do not let the conclusion balloon — compactness is part of the genre.
Step-by-step: how to write a research paper conclusion
1. Restate the thesis in new words
Return to the central argument — but do not copy the wording of the introduction. The reader has read the paper; they have earned a restatement that reflects what has been shown. "This paper has argued that..." is acceptable as a signpost, but the claim itself should be worded with the precision the body has earned. If the thesis shifted during drafting, match the restatement to the argument the paper actually made.
2. Synthesize rather than summarize
Do not walk section by section through the paper. Combine. If your paper had three body sections, the conclusion should pull them into a single claim about what they collectively establish. "Section 2 showed X, Section 3 showed Y, Section 4 showed Z" is a summary. "Together, these analyses establish that X and Y are two faces of the same mechanism, one that Z complicates rather than contradicts" is synthesis.
3. State implications
Implications answer the so-what question. What does your argument change? Be specific about who acts differently and how. For theory: which existing account is revised, extended, or refuted? For practice: what should a clinician, teacher, policymaker, or practitioner do differently? For the field: what is the next step in the conversation? Avoid generic claims ("this research has important implications"); name the specific claim and the specific actor.
4. Acknowledge limitations honestly
Every paper has limits. Name the two or three that most affect interpretation — sample composition, scope of the argument, methodological choices, time window of the data. Do this honestly, not apologetically. "The findings are based on a convenience sample of U.S. undergraduates, which limits generalization to other populations" is honest and useful. "This paper was limited by time and space and could not address..." is throat-clearing. Honest limits build credibility; they do not undermine it.
5. Point to future work
Identify the specific next question the paper opens. Make it concrete enough that another researcher could act on it. "Future work should extend these findings to adolescent populations using longitudinal designs" is actionable. "Further research is needed" is filler. The point-to-future sentence demonstrates that you understand where the paper sits in a larger trajectory.
6. Close with a sentence that lands
The final sentence should have weight. Return to the opening image, invoke the implication one more time, or deliver a compact restatement with a strong last word. Avoid trailing off with "therefore, more research is needed" — it is the conclusion-to-the-conclusion that readers remember, and "more research is needed" is forgettable.
7. Do not introduce new evidence
If a citation, a dataset, a case, or an argument is central enough to include in the conclusion, it needs to live earlier in the paper where you can develop it. The conclusion interprets; it does not add. Catching yourself introducing a new source in the conclusion is a signal to move that source into the body and rewrite accordingly.
Stuck at the start? PaperDraft scaffolds a starting conclusion — the moves, the register, the structure — for you to revise. Start this paper — free.
Example conclusion
Below is the conclusion to the hypothetical sociology paper used in our introduction guide, annotated in brackets.
[Thesis restated in new words, informed by what the body showed] The analyses above have traced how the control test, as applied in recent U.K. and Californian judgments, does in fact register the algorithmic dimension of platform management — more fully than either party in those cases conceded — while systematically understating the economic-dependence conditions under which drivers work. [Synthesis across sections, not section-by-section] Together, the doctrinal history and the comparative reading of judgments against empirical driver accounts suggest that the tests currently in use are neither as obsolete as platform critics contend nor as adequate as platform defenders claim. The problem is not the concept of control but its distribution: algorithmic control is legible to courts; economic dependence is less so.
[Implications, specific actor and action] For labor scholars, this points to a rebalancing of analytical attention toward the economic-dependence dimension, which has received less empirical treatment than algorithmic control in the past five years of platform-labor literature. For courts considering driver classification, the finding suggests that existing doctrine can accommodate the relevant facts if the economic-dependence prong is given evidentiary weight comparable to the control prong.
[Honest limitations] The comparison here is limited to three judgments in two jurisdictions and to English-language empirical accounts of driver work. Jurisdictions with different doctrinal starting points — Germany's Arbeitnehmerähnliche Personen category, for instance — may yield different alignments. The empirical base, drawn primarily from ride-hailing, may not extend to delivery and other platform sectors with different control architectures.
[Future work, concrete next question] The most productive next step would be a cross-sectoral comparison that held the doctrinal test constant across ride-hailing, delivery, and domestic-platform work, testing whether the alignment identified here is specific to ride-hailing or generalizes to platform labor more broadly. [Closing sentence with weight on the last word] Whether the control test survives the coming decade of platform-labor litigation will depend less on its conceptual adequacy than on whether courts are willing to read economic dependence as carefully as they now read control.
Notice what is not there: no summary of each section, no new citations, no apology for scope. Every sentence is doing one of the five moves.
Common mistakes
- The summary conclusion. Walking section by section through the paper you just wrote. Readers remember what they read. Synthesize instead.
- The new-evidence conclusion. Introducing a citation or a case in the final paragraph leaves it undefended. If it belongs in the paper, move it into the body.
- The apologetic conclusion. "Due to the limited scope of this paper..." is a weak frame. Name limits as honest scope decisions, not failures.
- The "more research is needed" fade. Generic call for future work that could be pasted into any paper. Name the specific next question.
- The verbatim repeat of the introduction. The conclusion is not the introduction with different punctuation. The reader has done the work of reading the paper; reward that with a restatement that has been through the argument — and with the honest attribution standards our academic responsibility guide describes throughout the paper.
How PaperDraft helps you start
PaperDraft is a writing assistant, not a paper generator — the draft is your starting point, not your submission. For research paper conclusions, PaperDraft scaffolds a starting structure with the five moves in place — restatement, synthesis, implications, limitations, future work — and a closing-sentence slot. You rewrite the scaffolded restatement to match the argument your paper actually made, supply the specific implications and limits only you know, and land the final sentence in your own voice. Start scaffolding at PaperDraft's research-paper workflow, and see the companion introduction guide for how the opening funnel sets up the moves the conclusion echoes in reverse.
Frequently asked questions
How long should a conclusion be?
Typically 5–10% of the paper. A 3,000-word paper: 150–300 words. A 6,000-word paper: 300–600 words. A thesis chapter: a full section. Longer than 10% usually means the conclusion has drifted into summary or repeated discussion.
Can I introduce a new idea in the conclusion?
No. Interpret what you have already shown; do not add. If an idea is important enough to appear in the conclusion, it belongs earlier in the paper where you can develop and defend it.
Should I repeat my thesis in the conclusion?
Restate, do not repeat. The reader has read the paper — the thesis should be restated in language informed by what the body has established, not in the exact words of the introduction.
Do I have to include limitations?
In empirical work, usually yes — reviewers and graders expect them. In argumentative or analytical work, limits may be framed as scope decisions rather than methodological limits, but naming them still strengthens the argument. Honesty here helps you rather than hurts.
Can I use a drafting tool to help with my conclusion?
Many programs allow scaffolding with disclosure, provided the final writing is yours. The conclusion in particular benefits from your own judgment — it is the place where you, not a tool, decide what the paper has shown. See our AI disclosure guide for program conventions and honest disclosure wording.