How to write a literature review, step by step

Many reviews go wrong before the first paper is opened. Without a clear scope, every paper looks relevant and the reading never ends.

Start by writing your review question in one sentence. It should name what you are studying, the effect you care about and the setting.

Then set the boundaries. Decide what is in and what is out, and write down why.

• Topic: ordinary concrete only. Fly ash in geopolymer or self-compacting concrete is left out.
• Time: studies from roughly the last fifteen years, plus a few older papers that everyone cites.
• Type: laboratory and field studies that report measured results.
• Language: English, because that is what he can read and check.

These boundaries may shift a little as you read, and that is fine, as long as every change is a decision you made and noted. The same note later becomes the scope paragraph at the start of your review.

A systematic search is one that someone else could repeat and get roughly the same results. That needs three things: the right databases, a written search string and a log.

Pick two or three databases that cover your field well. Varun used Scopus through his college library, Google Scholar, and ScienceDirect, which hosts several of the main concrete journals.

Next, list the key ideas in your question and every word authors use for each one. Fly ash, for example, is also called pulverised fuel ash, the term used in British and Indian standards. Then join the words with Boolean operators.

Record every search in a small log: the date, the database, the exact string, any filters and the number of results. When your examiner asks how you found your papers, the log is your answer. How to search for research papers like a pro covers databases, keywords and Boolean operators in more detail.

Finally, check the reference lists of the best papers you find, and the newer papers that cite them. This often turns up important studies that your keywords missed.

Varun's searches returned more than four hundred results. He could not read them all, and he did not need to. Screening is how you decide, fairly and quickly, which papers stay.

Write your rules before you start, so that you judge every paper the same way. These rules are called inclusion and exclusion criteria.

Screen in two rounds.

1. Titles and abstracts.Read only these, and sort each paper into keep, set aside or not sure. Keep the not sures for now.
2. Full texts.Open the papers that are left and check them against the same rules. Write a short reason for every paper you set aside.

Varun went from 412 results, to 318 after removing duplicates, to 74 after the first round and 38 after the second. If you are writing a systematic review, guidelines such as PRISMA ask you to report numbers like these in a flow diagram. For a thesis chapter a sentence or two is usually enough, but keep the numbers anyway.

Now read the papers that stayed. They do not all need the same care: a quick first pass tells you which ones deserve a slow, full reading. How to read a research paper shows a three-pass method that saves hours.

As you read, do not scatter notes across the margins of sixty PDFs. Fill in a table instead, with one row per paper and one column per question you ask of every paper. This is called a literature review matrix.

The rows above are invented to show the idea. Varun's real matrix had 38 rows, with columns for the mix, the test ages, the strength results, the durability tests and the limitations. It took him two weeks to fill, and it was the most useful thing he made.

With a full matrix you can read down a column and see at once where studies agree and where they do not. This guide to the literature review matrix explains how to set one up and which columns to use.

Once the matrix is full, patterns start to show. Several papers report the same effect, a few disagree, and one question keeps coming up that nobody has settled. These patterns are your themes.

Varun wrote each main finding on a sticky note and moved the notes around his desk until they settled into groups. Four themes came out: early strength, long-term strength, durability, and how much the results depend on the fly ash used.

Then choose how to organise the chapter. There are four common ways.

Many reviews are thematic, sometimes with a short chronological note at the start of a theme. Varun's outline followed his four themes, added a short section on test methods, and ended with a section on gaps.

This is the step that turns a list of papers into a review. Each paragraph should make one point supported by several sources, instead of describing one paper.

A simple recipe for each paragraph:

1. Topic sentence.State the point in your own words, as a claim.
2. Evidence.Bring in two to four sources that support or test the claim.
3. Compare.Say where the sources agree, where they differ and why they might differ.
4. So what.End with what this means for the field or for your own question.

The studies in the example are invented, but the shape is what matters. The first sentence makes a claim, the middle brings in several sources and explains a difference, and the last sentence points to a gap. Paragraph by paragraph, the review builds towards your own research question.

Write in your own words throughout. Quote only when the exact wording matters, and cite every idea that came from someone else.

Leave the first draft alone for a day or two, then read it the way your examiner will. Check each of these:

• Every paragraph starts with a claim, not with an author's name.
• Every claim has a source, and you have checked that each source says what you say it says.
• Recent work is included. Run your searches again near the end, because new studies keep appearing.
• Citations are complete and in one style, the one your university asks for. A reference manager such as Zotero or Mendeley saves hours here.
• The text is in your own words. If your university runs a plagiarism check, run this chapter through it too.
• The review ends with a clear gap that leads to your research question.

Then ask your guide or a classmate to read it. Varun's first draft came back with red ink on most pages, but the structure held, and the second draft took days instead of weeks.

Tools can shorten the early steps. Fiberarticle's Literature Reviewer searches arXiv, OpenAlex, Semantic Scholar and Crossref, screens papers against your inclusion and exclusion criteria, reads the open-access full texts, and builds an evidence matrix you can export as CSV, along with a draft review in which every point is tied to a numbered source. Treat it as a head start, not a finished chapter: it cannot read papers behind paywalls, and you still need to read the key papers yourself and write the argument in your own words.