How to structure a journal article: a section-by-section guide with word counts

A full journal article is usually 8,000 to 11,000 words, or about 24 to 26 printed pages, plus 60 to 120 references. This guide breaks that length down section by section: what each section must do, how long it should be, and the mistakes reviewers punish in each one.

Writing a journal article is easier when you know its shape before you start. Journals in management, business, psychology, education and the wider social sciences, published by houses such as Emerald, Elsevier, Wiley, SAGE and MDPI, follow a remarkably similar structure. We studied full-length articles from all five while building the journal article format in Soba Researcher. Below is the structure they share, with practical word counts for an article of about 10,000 words.

The structure at a glance

SectionTypical lengthWhat it must do
Title, abstract and keywords250 to 300 wordsTell the reader the purpose, method, findings and value in one paragraph or a structured block
Introduction900 to 1,200 wordsSet the problem, the gap, the aim, the research questions and the contributions
Literature review2,200 to 2,800 wordsSynthesise prior work by theme and end with a precise research gap
Theoretical framework and hypotheses900 to 1,200 wordsExplain the theory and argue each hypothesis or proposition from it
Methodology1,200 to 1,600 wordsDescribe the design, sample, data, measures and analysis well enough to repeat
Findings or results1,600 to 2,200 wordsReport the evidence for each research question, without discussing the literature at length
Discussion1,500 to 2,000 wordsInterpret the findings against prior work and give theoretical and practical implications, limitations and future research
Conclusion400 to 600 wordsAnswer each research question and state the main message
References60 to 120 sources, 4 to 8 pagesList every work cited, in the journal's style

Most articles also carry four to six tables and three to five figures. A literature synthesis table, a conceptual framework figure, a sample profile table and a table of future research directions appear again and again.

1. Title, abstract and keywords

Many management and business journals, Emerald's among them, ask for a structured abstract. It uses fixed headings, and it is a useful discipline even when a journal does not require it:

  • Purpose: what the study sets out to do, and why.
  • Design/methodology/approach: the design, sample, data and analysis, in one or two sentences.
  • Findings: the main results, stated specifically.
  • Research limitations/implications: what limits the study and what follows for future research.
  • Practical implications: what managers, practitioners or policy makers should do differently.
  • Originality/value: what is new and why it matters.

Follow the abstract with five to seven keywords and, where the journal asks for it, the paper type (research paper, conceptual paper or literature review). Choose keywords that people actually search for, because they decide whether your article is found.

Common mistake: an abstract that describes what the paper will do ("this paper discusses...") instead of what it found. Every claim in the abstract must be supported later in the findings.

2. Introduction

A strong introduction moves from the broad to the specific in eight to ten paragraphs. Open with the context and why the topic matters now. Narrow to the problem, then to what existing research has and has not done. State the aim, then three or four numbered research questions. Say plainly what the study contributes ("This study makes three contributions..."). Close with one short paragraph that outlines the rest of the article.

Common mistake: an unclear contribution. In our analysis of 2,378 peer reviews, an unclear or incremental contribution was raised in 22% of reviews on rejected papers, against 16% on accepted ones. Make the contribution impossible to miss.

3. Literature review

The literature review is usually the longest section. Organise it into four or five themes, each under its own subheading, and synthesise rather than summarise: say where studies agree, where they conflict and why. A table that sets out 10 to 14 key studies (author, context, method, key finding, limitation) helps reviewers see the field at a glance. End with a short subsection called Research Gap that states exactly what is still unknown and how your study addresses it.

Common mistake: an outdated or thin review. Cover both the foundational studies and the work of the last three to five years, and read what you cite. Most paragraphs in a good review cite two to four sources.

4. Theoretical framework and hypotheses

Choose one or two established theories that genuinely fit the problem, and explain how they apply. Then argue each hypothesis (for quantitative work) or proposition (for conceptual or qualitative work) from the theory and the literature, one per paragraph, numbered H1, H2 and so on. Finish with a conceptual framework figure that shows the constructs and the hypothesised links between them.

Common mistake: hypotheses that appear without an argument. Unjustified assumptions were one of the criticisms most strongly linked to rejection in the reviews we analysed: 26% of reviews on rejected papers against 18% on accepted ones.

5. Methodology

The methodology must let another researcher repeat the study. Use subsections: research design and philosophy; context and sample; data collection; measures or coding procedure; data analysis; rigour (validity and reliability, or trustworthiness for qualitative work); and ethical considerations. Justify every choice, from sample size to analytic technique, and cite the methodological sources you follow. A figure of the research design and a table of the sample profile are standard.

Common mistake: missing detail. If a reader cannot tell how participants were recruited, what was asked or how the data were coded, the reviewer will ask.

6. Findings or results

Report the evidence in the order of your research questions or hypotheses, one subsection each. For quantitative work, give descriptive statistics, the exact tests, effect sizes and confidence intervals. For qualitative work, give each theme with representative quotes and a short interpretation. Keep discussion of the literature for the next section. Two tables and one or two figures are typical here.

Common mistake: figures and tables that do not explain themselves. Refer to each one in the text before it appears, label every axis and unit, and make sure the numbers match the text exactly.

7. Discussion

Most journals expect four parts. Discussion of key findings interprets each result against prior studies, saying where it confirms, extends or contradicts them. Theoretical implications states what the study adds to the theories it used. Practical implications gives concrete actions for practitioners and policy makers. Limitations and future research is honest about the study's weaknesses and sets out a research agenda, often as a table of directions, example questions and suggested methods.

Common mistake: weak comparison with prior work. Compare your findings with named, recent studies, not with the literature in general.

8. Conclusion

Answer each research question in turn and state the main message in a few paragraphs. Do not introduce new citations, results or claims. Many journals then ask for declarations: author contributions, funding, conflicts of interest and a data availability statement.

9. References: how many, and how to get them right

Full journal articles in these fields commonly cite 60 to 120 sources, which takes four to eight pages in a single-column layout. Use the style the journal asks for; APA 7th edition, with author-date citations in the text, is the most common in the social sciences. Every source must be real, and must support the claim it is cited for. Fabricated references are one of the fastest ways to lose a reviewer's trust, and they are a known weakness of general-purpose AI writing tools.

A quick checklist before you submit

  • Does the abstract follow the journal's format, and is every claim in it supported later?
  • Are the research questions numbered and answered, in order, in the findings and the conclusion?
  • Does the literature review end with a precise research gap?
  • Is every hypothesis argued from theory, and shown in a framework figure?
  • Could another researcher repeat the study from the methodology alone?
  • Is every figure and table referred to in the text, with numbers that match?
  • Does the discussion compare findings with named prior studies?
  • Is every reference real, correctly formatted and cited for a claim it supports?

Drafting a journal article with Soba Researcher

Soba Researcher has a journal article format built on this structure. From a title, or from your own dataset, it drafts a full article of about 24 to 26 pages in eight steps: structured abstract and introduction, a two-part literature review with a synthesis table, theoretical framework with a conceptual figure, methodology, findings, discussion and conclusion. Its references are real published papers found in OpenAlex, cited in the text in APA style. A panel of 120 AI Review Agents then critiques the draft against the mistakes described above, and the article is rewritten to answer them.

Treat the result as a first draft to check and develop, not a finished submission: verify every claim and reference, carry out the research the paper describes, and follow your journal's rules on disclosing AI use. Read more in our FAQ, or try it free at sobaresearcher.com.

Key takeaways

  • A full journal article is typically 8,000 to 11,000 words (about 24 to 26 pages) with 60 to 120 references.
  • The literature review and the findings are usually the longest sections; the conclusion is the shortest.
  • A structured abstract (purpose, design, findings, limitations, practical implications, originality) is standard in many management journals.
  • Most reviewer criticisms are about explanation, not ideas: state the contribution, justify every choice, compare with named studies, and make every figure and reference count.

Word counts are typical ranges for full-length articles in management and social-science journals, drawn from published articles and from our study of public peer reviews. Always follow the specific author guidelines of the journal you submit to; they take precedence over any general guide.