Skip to content
TAtaimoorasghar.com

November 4, 2026 · 17 min read

What Are the 7 Steps in Writing a Literature Review?

What are the 7 steps in writing a literature review? Define, search, screen, organise, synthesise, analyse and write, with a checklist for each step.

What are the 7 steps in writing a literature review? In the sequence set out by the University of Texas at Austin Libraries, they are: define the research question; determine inclusion and exclusion criteria; choose databases and conduct the search; review your results; synthesise the information gathered; analyse the information gathered; and write the literature review. That list is a working order, not a one way street. In practice you will loop back, especially between searching, screening, and synthesis, as the literature sharpens your question and your question sharpens your search.

This guide walks through each step as a doctoral writer actually meets it, with the decision you have to make at each stage and the mistake that most often wastes time there. It is written for a PhD thesis chapter, but the same seven moves work for a standalone review, a proposal chapter, or a journal review article, with the depth adjusted to the job. If you have already worked through the framework and length guides on this site, this is the process piece that sits between them: the framework tells you what good synthesis looks like, the length guide helps you budget the words, and these steps tell you how to get from a blank page to a defensible chapter.

What a literature review is, and what it is not

A literature review is a critical account of existing research on a topic, organised so that it answers a question. It shows what is known, where studies agree and disagree, how strong the evidence is, and where a gap remains that your study can fill. It is not a summary of everything you have read, and it is not an annotated bibliography with the annotations joined together. A bibliography lists sources. A review builds an argument from them.

The distinction matters because it changes how you read. If you are only summarising, you can stop at what each paper found. If you are reviewing, you have to ask a second set of questions about every paper: what method was used, how large and how representative the sample was, what the authors could not show, and whether another study reaches a different conclusion from similar data. The University of Texas at Austin guide makes the same point in its synthesis and analysis steps, asking writers to evaluate strengths and weaknesses, bias, methodologies, and thoroughness, and then to group results into a structure that supports the research question. Keep that test in mind from step one. It is what stops a long chapter from reading like a list.

Step 1: Define the research question

Start with a question that is specific enough to search and open enough to be worth a review. A topic such as “mental health in doctors” is too broad to search well. It will return thousands of papers across unrelated populations, settings, and outcomes, and you will spend weeks deciding what counts as relevant. A question such as “what does the literature say about burnout interventions for junior doctors in hospital settings” is searchable: it names a population, a phenomenon, and a setting, and it already suggests keywords.

Write the question down, then write a scope sentence beside it. The scope sentence should state the central topic, the population or context where relevant, the main concepts, and the purpose of the review inside your thesis. If a time period or a geographical boundary matters, state why it matters rather than adding it by habit. A scope sentence is not a promise you can never change. It is a filter for the first search, and you should expect to revise it once you see what the literature actually contains. The University of Texas at Austin guide notes that exploratory searching may be needed at this stage to get a sense of scope and to decide whether to narrow or broaden the focus. That is normal. A question that survives first contact with the literature unchanged is rare, and a question that never changes is usually a sign that the search was too narrow.

Two practical habits help here. First, keep a one page working note with your question, your scope sentence, and the terms you have tried, so that a supervisor can see your thinking without sitting through a retelling of every search. Second, agree the question with your supervisor before you build a large search. An early conversation about scope saves far more time than a late conversation about a chapter that covers the wrong field.

Step 2: Determine inclusion and exclusion criteria

Before you run a full search, decide what will count as in and what will count as out, and write those rules down. Inclusion and exclusion criteria are the second step in the University of Texas at Austin sequence, and they do the quiet work of keeping a review honest. Without them, it is easy to include a striking paper because you liked it and to leave out an awkward paper because it complicated your argument. With them, every borderline decision has a reason you can explain to a reader or an examiner.

Typical criteria cover study type, population, setting, language, and date range, but only where each criterion has a reason tied to your question. A date limit needs a justification, such as a policy change, a new method, or a field that began at a known point. A language limit needs the same honesty: if you can only read sources in one language, say so and note it as a limitation rather than pretending the field stops at that language. The same applies to study design. If your question is about lived experience, qualitative studies belong in the review even if they do not carry the largest sample sizes. If your question is about effectiveness, study design and risk of bias belong near the top of your appraisal.

Keep the criteria short enough to apply consistently. A list of twenty rules that you cannot remember while screening will not be applied, and a review that applies different rules on different days is hard to defend. Five or six clear rules, written down before the main search, are worth more than a long list written after the results are in.

Step 3: Choose databases and conduct the search

Choose databases for your field, not for habit. A medical question needs different databases from a humanities question, and a question that spans two fields may need both. Ask your subject librarian which databases index the journals that matter in your field. The University of Texas at Austin guide makes this point directly, noting that the first four steps are the best points at which to contact a librarian, who can help you determine the best databases, assess scope, and formulate a search strategy. That advice is worth taking literally. A thirty minute conversation with a librarian can prevent a search that misses a whole body of work because it used the wrong index terms.

Build your search from concepts, not from single words. Break your question into two or three concepts, list synonyms and related terms for each, and combine them with AND between concepts and OR within a concept. Use controlled vocabulary where your database offers it, because authors do not always use the words you would choose. Keep a search log from the first run: the database, the date, the exact search string, the number of results, and any limits applied. That log is not bureaucracy. It lets you rerun a search months later, it lets a supervisor check your coverage, and it gives you the methods paragraph of your chapter almost for free.

Search more than one database, and follow references in both directions. The University of Texas at Austin guide recommends examining the citations of relevant articles for keywords, authors, and previous research, and using cited reference searching to find later work that cites a key paper. That forward and backward checking often finds the papers a keyword search misses, including older foundational work and newer work that uses different terminology. Record those additions separately in your log, so that your search remains reproducible even when it includes hand searching.

Step 4: Review your results

Screening happens in two passes. First, screen titles and abstracts against your inclusion and exclusion criteria, and set aside anything that clearly fails. Second, read the full text of what remains and apply the criteria again, more strictly. Many papers look relevant in an abstract and fail on full reading because the population, the intervention, or the outcome is different from your question. Keep a short reason for each full text exclusion. You do not need a formal flow diagram for every thesis chapter, but a clear account of how many records were screened, how many were read in full, and why the rest were excluded makes the chapter far easier to trust.

Save results in a reference manager from the start. The University of Texas at Austin guide names Zotero, Mendeley, and EndNote as examples. Whichever tool you use, use it consistently, store the full text or a stable link where your licence allows, and add your own notes to each record as you read. Notes written on the day you read a paper are worth far more than notes reconstructed months later from a highlighted PDF. For each paper, record the question, the method, the main finding, the limitation that matters for your review, and one sentence on how it connects to your question. That five line note is the raw material for synthesis.

Do not try to read everything in depth at this stage. The purpose of screening is to separate the papers that will shape your argument from the papers that are background only. Background papers still get cited where they establish context, but they do not all need a paragraph of critical appraisal. Reading everything at equal depth is the most common cause of a literature review that never gets written.

Step 5: Synthesise the information gathered

Synthesis is where a review stops being a list and starts being an argument. It means grouping sources by what they say together, not by who wrote them. The University of Texas at Austin guide asks writers to group results into an organisational structure that will support why the research needs to be done, or that provides the answer to the research question. In practice, that structure is usually thematic: studies are grouped by a concept, a mechanism, a population, or a debate, and each theme is discussed as a conversation between studies.

A concept matrix helps here, even a simple one. Put themes across the top and papers down the side, and mark where each paper speaks to each theme. Patterns appear quickly: a theme where every study agrees, a theme where two methods give different answers, a theme where the evidence thins out. Those patterns become your section headings and your topic sentences. Without a matrix or an equivalent map, it is easy to organise by author instead, which produces the familiar paragraph per paper structure that examiners criticise.

Synthesis also means naming the relationship between studies in plain language. Do studies agree, and if so, on what exactly? Do they disagree, and is the disagreement about findings, methods, populations, or definitions? Does one study explain another, qualify it, or contradict it? A sentence that says “three studies found X, while two studies using a different measure found Y” does more work than three separate summaries. It also prepares the ground for step six, because disagreement is often a clue about method or measurement rather than a simple conflict.

Organise thematically or methodologically rather than chronologically, unless chronology is itself your argument, for example when you are showing how a field changed after a landmark study. A chronological walk through the literature can be useful as a first draft, but it should rarely survive as the final structure. Readers remember themes. They rarely remember a sequence of publication years.

Step 6: Analyse the information gathered

Analysis is the critical judgement that sits behind synthesis. Synthesis groups the evidence. Analysis weighs it. For each theme, ask how strong the evidence is, what its limits are, and whether the conclusion you want to draw is supported or only suggested. The University of Texas at Austin guide lists the relevant checks as strengths and weaknesses of individual sources, bias, methodologies, and thoroughness. Apply those checks at the level of the theme, not only at the level of the single paper, because a theme built on five small studies with the same flaw is not stronger than one good study simply because it has more citations.

Be specific about flaws. A vague sentence such as “these studies had limitations” tells an examiner nothing. A specific sentence such as “all three studies relied on self report without an objective measure, and none adjusted for prior experience” tells the reader exactly how far the finding can be trusted. The same discipline applies to strengths. Note where a study used a large sample, a validated measure, a comparison group, or a design that fits the question, and say why that matters for your theme.

Analysis is also where you find the gap honestly. A gap is not simply a topic nobody has studied. It can be a population that has not been studied, a method that has not been tried, a contradiction that has not been resolved, or a theory that has not been tested in a new setting. State which kind of gap you are claiming, and show the evidence for it from the synthesis. A gap paragraph that follows from the themes you have just discussed is persuasive. A gap paragraph that appears from nowhere, after a list of summaries, is not.

Step 7: Write the literature review

Writing is the seventh step, but it should not be the first time you put sentences together. If you have kept short notes for each paper and a map of your themes, the draft is mainly a matter of turning those notes into connected prose. Start with an introduction that states the question, explains why the review matters, and previews the themes in the order you will discuss them. Then take one theme per section, open each section with a topic sentence that states what the theme shows, discuss the studies as a group, and close the section with an implication for your research question. End with a conclusion that summarises what is known, what is contested, and what gap your study addresses.

Write in your own critical voice, and let citations support your sentences rather than replace them. A paragraph that begins with an author name and ends with a finding, repeated ten times, is a list in paragraph form. A paragraph that begins with a claim about the theme, supports it with two or three citations, notes a disagreement, and ends with your judgement, is a review. Use signposting sparingly but clearly: a sentence at the start of a section that tells the reader what the section will show, and a sentence at the end that tells the reader what follows from it.

Revise in passes, not in one read through. First, check structure: does each section serve the question, and does the order build an argument? Second, check synthesis: is every section a conversation between studies rather than a sequence of summaries? Third, check accuracy: are findings, sample details, and attributions correct, and is every citation in the reference list? Fourth, check language: have you removed repetition, tightened long sentences, and kept the tone critical without being dismissive? Dakota State University Library’s guide to the literature review process makes a useful parallel point, ending its six step sequence with evaluating your work as a final check that it is complete and of high quality. Treat that final evaluation as a step in its own right, not as an afterthought.

Putting the seven steps together

In real doctoral writing, the steps overlap. A search will reveal a subfield you had not considered, and you will return to step one to narrow or broaden the question. Synthesis will reveal that two themes are really one, and you will return to step four to recheck a set of papers. Analysis will reveal that a key study is weaker than you first thought, and you will return to step three for a targeted search on that point. That looping is not failure. It is how a review becomes defensible, provided each loop updates your search log, your criteria, and your theme map rather than happening only in your head.

A simple checklist keeps the process visible. Before you call a draft finished, confirm that you can answer yes to each of these: the question and scope are written down; the inclusion and exclusion criteria are written down and were applied consistently; the databases, search strings, and dates are logged; screening reasons are recorded for full text exclusions; themes are organised by concept rather than by author; each theme weighs strengths, weaknesses, and bias; and the conclusion states a specific gap that follows from the synthesis. If any answer is no, that step is not finished, however polished the prose looks.

Time planning follows the same logic. The early steps, defining the question, setting criteria, and searching, feel slow because they produce no prose, but they determine whether the prose you later write is about the right field. The middle steps, screening and synthesis, are where most doctoral writers stall, usually because they are reading at equal depth or organising by author. The writing step then feels hard because the thinking has not been done yet. If writing feels hard, the usual remedy is not more writing time. It is a clearer theme map, a sharper question, or a second pass through screening with stricter criteria.

Common mistakes, and how each step prevents them

The most common mistake is starting with a search instead of a question. It produces a large pile of papers and no way to decide which ones matter. Step one and step two prevent it, provided the question and criteria are written down before the main search. The second mistake is treating every paper equally. It produces a chapter that is long, flat, and hard to read. Steps four and six prevent it, by separating background reading from argument shaping papers and by weighing evidence at theme level.

The third mistake is organising by author. It produces a chapter that reads as “Smith found, Jones found, Patel found” and never reaches a judgement. Step five prevents it, if you build a theme map before you draft. The fourth mistake is claiming a gap without evidence. It produces a conclusion that an examiner can dismantle in one question. Steps five and six prevent it, because a gap that follows from synthesis and analysis can be shown, not merely asserted. The fifth mistake is leaving referencing and record keeping to the end. It produces missing citations, duplicated references, and a final week spent reconstructing a search you cannot remember. Steps three and four prevent it, if the log and the reference manager are used from the first search.

Frequently asked questions

Do the 7 steps have to happen in strict order?

No. The sequence is a working order, and looping back is expected. Searching can refine the question, synthesis can trigger a targeted search, and analysis can send you back to screening. What matters is that each loop updates your written question, criteria, search log, and theme map, so the final chapter can show its working.

How is a literature review different from an annotated bibliography?

An annotated bibliography describes sources one by one. A literature review groups sources by theme, compares and contrasts their findings, weighs their methods, and builds an argument about what is known and what remains unknown. If your draft could be rearranged by author without losing meaning, it is still closer to a bibliography than to a review.

How many sources should a PhD literature review include?

There is no universal number, and any guide that gives one without reference to scope, field, and thesis structure should be treated with caution. The right number is the number needed to cover your themes, represent the main debates, and support the gap claim, without padding. Coverage and saturation matter more than a count: when new papers stop changing your themes, your search is approaching completeness for that scope.

How long should a PhD literature review be?

Length varies by discipline, thesis structure, and university rules, and it should be budgeted separately from the process described here. The companion guide on this site, How Long Should a PhD Literature Review Be, sets out planning ranges and how to adjust them for your field. Agree the length with your supervisor early, so that the seven steps are scoped to the chapter you actually have to write.

Can I use the 5 C’s framework inside these 7 steps?

Yes. The 5 C’s, cite, compare, contrast, critique, and connect, are most useful inside steps five, six, and seven. Use them as a paragraph level test: does each theme cite the key studies, compare and contrast their findings, critique their methods, and connect the result to your research question? The companion guide on this site, What Are the 5 C’s of a Literature Review, explains that framework in full.

Should I ask a librarian for help?

Yes, and early. The University of Texas at Austin guide identifies the first four steps as the best points to contact a librarian, for help with databases, scope, and search strategy. A librarian can also help with controlled vocabulary and with rerunning a search cleanly, which saves time later when you need to update the review before submission.

Related reading

What Are the 5 C’s of a Literature Review?
How Long Should a PhD Literature Review Be?
What Are the 5 Components of an Abstract?
How Do You Respond to Peer Reviewer Comments?
How Can I Publish My Own Research Paper?

Sources

University of Texas at Austin Libraries, Steps in the Literature Review Process, Literature Reviews LibGuide: https://guides.lib.utexas.edu/c.php?g=1060589&p=7710319


Dakota State University Library, Graduate Research: Guide to the Literature Review, Literature Review

Steps: https://library.dsu.edu/c.php?g=22495&p=133184