How to write a literature review in business and management

In business and management, the literature review is where your theory comes from. Unlike fields where the review sets up an empirical question and then steps aside, a management review has to deliver a conceptual framework: the constructs you will measure, the relationships you expect between them, and the theoretical reason those relationships should hold.

That is why examiners are unimpressed by reviews that describe a lot of studies. The question they ask is whether the framework in your third chapter is visibly produced by the review in your second one. If a construct appears in your model without having been argued for in the review, it will be queried.

This guide covers the management-specific work: running a protocol-driven review, scoping with journal quality lists, building search strings from constructs, organising by concept rather than author, and turning the whole thing into hypotheses. The general AI-assisted workflow is in how to write a literature review with AI.

What a business and management literature review is graded on

Four criteria come up repeatedly in management marking rubrics, and they are worth writing on a card above your desk.

First, theoretical grounding: your review is organised around theories and constructs, not topics. Second, critical engagement: you evaluate the evidence rather than reporting it. Third, transparency: a reader can see how you found and selected the studies. Fourth, integration: the review produces something, usually a framework and a set of testable propositions.

The third criterion has hardened in recent years. Structured or systematic review methods have moved from being a specialist choice to being a default expectation in many management programmes, and even a narrative review is now expected to say where its sources came from.

Integration is what separates a pass from a strong chapter. A well-integrated review leaves the reader able to state, unprompted, what the field currently believes, where it disagrees with itself, and what your study will settle. If the reader finishes your chapter holding a list of studies rather than a position, nothing has been integrated.

Write the review protocol first

A protocol is a page that says what you will do before you do it. It costs an afternoon and saves weeks, mostly because it stops the search from expanding every time you find an interesting adjacent literature.

Your protocol should specify the following, and you should paste it more or less directly into your methods section afterwards.

• The review question, phrased so it can be answered rather than explored forever.
• Databases and platforms searched, with the date of each search.
• The exact search strings, including Boolean structure and truncation.
• Inclusion and exclusion criteria: date range, language, peer-review status, journal scope, methodology, industry or geography if relevant.
• How screening was done and by whom, and how disagreements were resolved.
• The data extraction fields, which become the columns of your matrix.
• How you will synthesise: thematic, theory-led, or a formal method such as meta-analysis.

Extract the same fields from every management study

Critical engagement is a marking criterion in nearly every management rubric, and it is the one students most often mistake for hedging. Criticism is not observing that a study had limitations. It is saying what those limitations mean for the claim you are trying to build.

The practical route is to extract the same fields from every paper, so that comparison is possible at all. Vary the fields by design where you must, but keep the core stable.

Protocol-driven reviews in management borrow their machinery from health research, so the reporting conventions will look familiar if you have ever seen a flow diagram of screening counts. The PRISMA walkthrough covers how those counts accumulate while you work, which is considerably easier than reconstructing them from browser history the week before submission.

The final item on the list below is where contribution often hides. Common-method bias in single-source survey data, samples drawn from one country's listed firms, or a construct measured with an instrument built for a different context are all standard weaknesses, and a review that names them precisely is already most of the way to a justified design.

• Theoretical lens, and the assumptions it quietly imports.
• Constructs, their stated definitions, and the instrument or items used to measure each one.
• Sample: size, source, level of analysis, country and sector.
• Design: cross-sectional, longitudinal, experimental, qualitative or archival, and whether causal language is warranted.
• Findings, including effect direction and any non-significant results the authors pass over quickly.
• Limitations the authors state, and the ones they do not.

Use journal quality lists to scope, not to judge

Management has more journal ranking infrastructure than most disciplines. The Chartered Association of Business Schools guide, the Financial Times list and the Australian Business Deans Council list are all used to define what counts as a quality outlet, and many supervisors will expect you to reference one of them when explaining your inclusion criteria.

Used well, these are scoping devices. Restricting to a defined tier is a defensible way to make an unbounded literature finite, and it is easy to state in your methods. Used badly, they become a proxy for judgement, and that shows: a review that includes a weak paper because it appeared in a highly ranked journal, or excludes the definitive study in your niche because its outlet is unranked, has stopped thinking.

The workable compromise is a tiered rule you state openly. Search comprehensively, apply a quality threshold for the core corpus, then admit specific outside sources where you can justify each one by name. That is honest and it protects you from the obvious question about the paper everyone in your field knows and you left out.

Remember that ranking lists are organised by discipline, not by question. A cross-disciplinary topic such as sustainability or digital work is covered in journals that sit outside the business lists entirely, and a threshold applied without thought will delete the most relevant half of your evidence. State the list you used, state the exception rule, and apply the exception where the topic demands it.

Build search strings around constructs, not topics

Topic searches return volume. Construct searches return relevance. If your question is whether transformational leadership improves team innovation in remote settings, do not search that sentence. Build one synonym block per construct and combine them.

The reason this matters more in management than elsewhere is that the same construct travels under several names across sub-disciplines, and adjacent constructs are often measured with overlapping instruments. Missing a synonym block means missing an entire research stream that has been asking your question under another label.

Run the search wide before you narrow it. Literature Finder reaches an internal corpus plus live OpenAlex, PubMed, Semantic Scholar and arXiv, which covers a great deal of management scholarship, and anything your library subscribes to separately can be added to the same project by uploading the PDF or pulling details from a DOI. Record the counts at each stage while you go, because reconstructing them later is miserable.

Keep a search log as a simple table: string, database, date run, records returned, records kept. It costs a minute per search and it is the difference between a methods section you write in an hour and one you rebuild over three days from memory and browser history.

Organise by concept, not by author

The concept-centric review is the standard the field expects, and the author-centric review is the one students hand in. If your paragraphs begin with author names and years, you are writing an annotated bibliography with the annotations run together.

The fix is mechanical. Build a matrix with studies as rows and analytical dimensions as columns, then write from the columns rather than the rows. For management work the useful columns usually include theoretical lens, constructs and how each was operationalised, sample and context, method and design, key findings, effect direction, and stated limitations.

Synthesis Lab builds that evidence matrix across the papers in your project, which turns the writing problem into a reading-down-a-column problem. When you can see that eleven studies used one instrument and four used another, and the four disagree with the eleven, you have a paragraph that is actually about something.

Once the matrix exists, draft each section from a single column rather than from the whole grid. One theme, one pass, one argument, then move on. Trying to write from the entire table at once is how reviews slide back into study-by-study description.

Name the gap precisely: theory, context or method

Vague gap statements are the single most common weakness in management dissertations. More research is needed is not a gap. A gap is a specific claim about what the existing literature cannot currently tell you, and it has a type.

Use the type to write the sentence, and your justification chapter almost writes itself.

• Theory gap: a relationship is documented but the mechanism explaining it has not been tested, or two theories predict opposite outcomes and no one has adjudicated.
• Context gap: the effect is established in one setting, sector, firm size or country and has not been examined where the boundary conditions plausibly differ.
• Method gap: findings rest on one design, such as cross-sectional self-report, and a different design would test whether they hold.
• Construct gap: a variable is measured inconsistently across studies, and the inconsistency may explain the mixed results.
• Population gap: the samples are drawn from a narrow group, such as one industry or one management level, and generalisation is unwarranted.

Turn the review into a conceptual framework and hypotheses

Every arrow in your conceptual model needs a paragraph in the review that earns it. Work backwards: draw the model, then for each path write down which studies support it, which theory explains it, and what remains uncertain. Any path without support is either an unjustified assumption or the contribution you need to argue for explicitly.

Hypotheses should be traceable to the same paragraphs. A clean pattern is to end each thematic section of the review with the proposition it produces, so the reader watches the model assemble rather than meeting it fully formed in the next chapter.

Be explicit about moderators and mediators. Much of the value in a management contribution is in boundary conditions, and reviews that flatten every finding into a main effect miss where the live argument is. If two literatures disagree, a moderator is often the resolution and often the thesis.

Keep the framework falsifiable. A model that predicts everything predicts nothing, and reviewers in management notice quickly when a set of hypotheses cannot fail. For each proposition, be able to say what result would count as disconfirmation, and check that the review has given a reason to expect the direction you predict rather than merely a reason to expect some association.

Citation style and reference hygiene

Most management programmes use APA or Harvard, both of which are among the twelve styles Reference Manager supports, and both punish inconsistency in author-date formatting more than students expect. Set the style once at the project level and let the bibliography derive from it rather than maintaining a list by hand.

Two hygiene rules matter in this field specifically. Never cite a foundational theory paper you have not read, because seminal work in management is heavily paraphrased and the received version often drifts from the original. And check that a construct definition you attribute to an author is actually theirs and not a later gloss.

• Every construct in your model is defined in the review, with a source.
• Every hypothesis maps to a section of the review that supports it.
• Search strings, databases, dates and screening counts appear in your methods.
• The gap statement names its type and cites the studies that leave it open.
• No seminal citation was inherited from a secondary source without reading the original.

What supervisors flag in a business literature review

The weaknesses supervisors flag in this chapter are consistent, and most of them are visible from the structure alone before anyone reads a sentence.

• An annotated bibliography in disguise: one paragraph per study, nothing connecting them.
• Constructs used interchangeably when the underlying measures differ, which manufactures agreement that is not there.
• Citing a theory from a textbook summary rather than the original, then building a hypothesis on the summary.
• A gap statement that says more research is needed without saying what that research would establish.
• A conceptual model containing paths that appear nowhere in the review.
• Recency without depth: forty citations from the last three years and nothing showing where the debate started.