Writing6 min readBy Saradha R, Senior Research AnalystPublished 2026-10-24

How to write a first draft of your research paper from your results

To write a research paper first draft, start with your results and methods, not the introduction. List the tables and figures, write what each shows, then write the methods that produced them, then the discussion, and last the introduction and abstract. A first draft only has to be complete and honest. Polish comes in later rounds.

Many researchers stare at a blank page because they begin with the introduction, the hardest part to write before the paper’s argument is clear. Working from the results outward gives you something concrete to write about at every step.

Why write the results first?

The results are the part you know best, and they decide what the rest of the paper must say. The methods must explain how you got them, the discussion must say what they mean, and the introduction must lead to the question they answer. If you write the introduction first, you often rewrite it when the results change.

IMRaD in one page

Write the results section from your tables and figures

1. List every table and figure you plan to include, in the order a reader should see them.
2. Under each, write one or two sentences stating what it shows, with the actual numbers.
3. Say which result answers which question or hypothesis.
4. Report weak and negative results too, and the baselines you compared against.
5. Keep interpretation for the discussion.

Template / example

Table 2 shows the error of each method on the held-out set. The proposed method has a lower mean error than the best baseline (0.31 compared with 0.36), and the difference holds across five random seeds (Figure 3).

The numbers above are an example of structure, not a real result. Your sentences should carry your own numbers.

Write the methods so someone can repeat them

• Describe the data: source, size, any cleaning or exclusions.
• Describe each method and baseline, with settings and versions.
• State how you evaluated: metrics, splits, seeds, statistical tests.
• Say what hardware and software you used.
• Cite the original source for each method you did not create.

Write the discussion without overclaiming

• Start with the main finding in one sentence.
• Compare with the closest papers you listed during the novelty check.
• Explain surprises and weak results honestly.
• State limitations plainly: data, scope, assumptions.
• End with what follows from the work, not with a list of everything you wish you had done.

A good test: for each sentence in the discussion, can you point to a result or a source behind it? If not, soften it or cut it.

Write the introduction last

Revise in passes

1. Check structure: does each section do its job?
2. Check claims: does each one have evidence or a source?
3. Check numbers and references against the originals.
4. Read it aloud or print it, to find unclear sentences.
5. Check the journal’s word limit and required statements.

Ask a colleague who did not work on the project to read it. If they cannot say what the contribution is, the paper is not clear yet.

A checklist before you share the draft

• Every table and figure is cited in the text and has a caption that stands alone.
• Every number in the text matches a table, figure or log.
• Every claim in the discussion points to a result or a source.
• Baselines and settings are described in the methods.
• Limitations are stated in plain words.
• References are complete and in one style.
• The title, abstract and keywords match the finished paper.

Write clearly, not impressively

Short sentences with one idea each are easier for reviewers and for readers whose first language is not English. Prefer the plain word, define terms once, and keep tenses consistent: past for what you did and found, present for what is generally true. Read the draft aloud to catch sentences that do not breathe.

Where a tool can help

In EasySeer, you choose the layout, citation style and tone, and decide whether to include an AI-use statement, a cover letter draft and a check of every claim against your results. It then writes the manuscript from your concept, code and results as DOCX and PDF, with a section outline beside the draft and a word count. The sample draft is 12,496 words.

It is a starting point, not a finished paper. Suggested references can be wrong or incomplete, so you must check them. You read the draft section by section, verify every number and reference and rewrite it in your own words, because you stay the author.

See the manuscript stageRead: draft your manuscript

Questions

Where should I start writing a research paper?

With the results and methods, because they are fixed by your work. Write the introduction and abstract last.

How long should a first draft be?

Long enough to include every result and argument. Cutting is easier than adding. Check the journal’s word limit at the revision stage.

Should I write the abstract first?

A short draft abstract can guide you, but the real one should be written last from the finished paper.

Can AI write my first draft?

A tool can produce a draft, but you are responsible for every claim and reference, and many journals ask you to disclose its use. See the post on using AI in research papers.

How do I know the draft is ready for co-authors?

When every table and figure is cited, every number matches the results, and each section does its job. Then let them mark it up.

Keep reading

Sources

Checked on 2026-10-24. Policies change, so read the current version before you submit.

• EQUATOR Network: reporting guidelines
• ICMJE recommendations
• COPE: Committee on Publication Ethics

Published 2026-10-24. Last updated 2026-10-24.