How to do a literature review: seven steps for the start of term

Term has started, your supervisor said “review the literature first”, and Google Scholar is showing you 4,000 results. A review is not reading a lot; it is deciding quickly what not to read and keeping the rest in order.

A literature review maps what has been said about your topic, with which methods, and where the gaps are. A thesis, a seminar paper or the introduction of an article all want the same map — only the scale differs.

1. Turn the topic into one question

“Social media and young people” is a topic; “How does short-video use affect the attention span of university students?” is a question. Searching before the question is sharp multiplies confusion, not results.

2. Write down keywords and their synonyms

At least two terms per concept: “attention span / sustained attention”. If you do not know the field’s vocabulary yet, the titles of the most-cited papers on our field pages hand it to you.

3. Pick your sources — in order

Google Scholar for breadth, national repositories for local-language work, subject databases (PubMed, IEEE Xplore, ERIC) for depth. We covered the first in the Google Scholar guide; for a paper you cannot access there is the open-access route.

4. Screen first, read second

Not every result gets read. Title → abstract → conclusion, two minutes each: is it relevant, is the method sound, where does the journal stand? Journal quartile and citation count are the ten-second signals for this pass; the predatory-journal checklist tells you what should not get through the door.

5. Snowball

When you find a good paper, go both ways: its reference list (backwards) and the papers citing it (forwards). That is how a field’s cornerstones surface — far faster than paging through search results.

6. Keep everything in one place, with a note

If your sources live in four tabs, two PDF folders and a chat message, there is no review. A reference manager (Zotero, Mendeley) or Pofolia’s lists: a list holds 500 papers, an account holds 10 lists, and each paper sits with its short summary and your note on phone and web alike.

7. Write the gap, not the list

The output of a review is not “I read these 40 papers” but “these three things are known, these two contradict each other, nobody has measured this”. A five-sentence summary per paper — template here — makes that sentence easy to write.

Pofolia’s AI summaries are the starting point of this step, not a substitute for the source: a summary rests on the abstract and text, so verify every claim that enters your thesis against the paper itself. We explain why in how to read AI summaries.

The three most common mistakes

  • Reading only the last two years: a field’s founding papers are usually older; “new” and “foundational” are separate lists.
  • Saving without reading: 300 PDFs and no memory of what any of them found. Every save deserves a one-sentence note.
  • Trusting a single database: Google Scholar is broad but has no quality filter; a national repository is essential for local work, a subject database for depth.

Blog