From search strategy to synthesis matrix to a structured draft — a repeatable workflow for literature reviews that examiners actually enjoy reading.
A literature review is not a list of summaries. It is an argument about what a field knows, how it knows it, and where the gaps are — built from other people's papers. This workflow takes you from a blank page to a structured draft in six repeatable steps.
This guide is for students facing the chapter or assignment that sounds simplest and stalls longest: 'review the literature'. It suits a dissertation chapter, a standalone review paper, or the background section of a project report. The workflow assumes no special software beyond a search engine, a reference manager and a spreadsheet or document for your matrix — though it works even better with them.
One sentence: what exactly are you reviewing the literature about, for what population, setting or system, and to answer what question? Every inclusion decision later gets easier with this sentence pinned above your desk. Reviews that skip this step collect papers the way magpies collect foil.
Here is a search string doing honest work, for a review on feedback timing in undergraduate science courses: ("formative feedback" OR "feedback timing" OR "delayed feedback") AND (undergraduate OR "higher education") AND (science OR STEM). Notice the pattern: synonyms for one concept grouped with OR inside brackets, concepts joined with AND, exact phrases in quotes. Paste it into Scholar for a first sweep, then adapt it to your subject database's own syntax — most databases have the same logic with slightly different symbols, and their help pages show the local dialect.
Decide your inclusion and exclusion rules before screening: date range, language, study types, settings. Apply them first to titles and abstracts, then to full texts. Keep a simple count of what fell out at each stage — that count becomes the methods paragraph of your review.
Create a table with one row per paper and columns for question, method, sample, key finding and limitations. Fill it as you read. Patterns that are invisible across twenty PDFs become obvious in one table: who agrees, who contradicts, which methods dominate, and which questions nobody has asked.
Add a final column: 'So what for my study?' If a paper cannot fill it in one line, it probably belongs in the background section, not the core review.
| Paper | Method | Key finding | Limitation |
|---|---|---|---|
| Study A | Survey of 212 undergraduates | Students preferred feedback within a week | Self-reported preference, not measured learning |
| Study B | Randomised trial, two cohorts | Delayed feedback improved retention on delayed tests | Single institution, one subject area |
| Study C | Interviews with 18 lecturers | Workload, not belief, drives feedback delays | Small sample; no student data |
An illustrative example with placeholder studies. Even in miniature the pattern jumps out: preference and learning point in different directions, and the lecturer study explains why practice lags both. That tension is a theme — and themes are what your review is made of.
One more writing habit separates synthesis from summary: the theme sentence. Every paragraph of a themed section should survive having its first sentence read aloud on its own — 'Evidence on timing is split between what students prefer and what testing shows.' If a paragraph's first sentence is a paper's name ('Smith et al. (2021) studied…'), you are summarising again. Useful stems: 'Across the quantitative studies…', 'Where X and colleagues found…, Y's team reported the opposite pattern…', 'The interview studies add a complication the trials cannot see…'
Group your matrix rows into three to five themes — by finding, method or debate, never 'Paper 1 says… Paper 2 says…'. Each theme becomes a section that opens with the overall picture, presents the strongest evidence on each side, and closes with what remains unresolved.
Draft your concluding gap paragraph first: what the field does not yet know, and why it matters. Then write each theme section so it points at that gap. A review structured this way reads as an argument with a destination — which is exactly what examiners reward.
Drowning in results (thousands of hits). Your concepts are too broad or your criteria too kind. Tighten in this order: add a date range, require the concept in the title or abstract, then narrow the population. Record each tightening in the search log — the funnel is a finding, not a failure.
Finding almost nothing. Work the synonyms harder (ask how the field itself names the thing — check the keywords of the two closest papers), remove one AND concept at a time, and snowball: the reference list of one good paper is a pre-screened library on your exact topic.
Themes will not emerge. Your matrix is probably sorted by paper. Re-sort it by finding — group the rows that agree, then the rows that contradict, then the rows using different methods. Themes are agreements, disagreements and methods wearing a trench coat; sorting exposes all three.
The draft reads like a shopping list of studies. Apply the deletion test: remove any paragraph that could not move to a different position in the chapter without breaking the argument. Survivors are synthesis; the rest is summary wearing synthesis's clothes — rewrite those around claims, not papers.
None of these steps requires talent, only sequence. Question, search, screen, extract, organise, argue: run the loop once and the next review takes half the time.
Long enough to establish the conversation and the gap — no longer. As a rough guide, in a dissertation the review chapter commonly runs at about a tenth to a fifth of the total length, but your institution's handbook and your supervisor's expectations outrank any rule of thumb.
There is no magic number, and markers can tell the difference between coverage and padding. A focused undergraduate review might rest on 25–40 well-chosen sources; a doctoral review may engage hundreds. Stop when new searches stop changing your themes — that point is called saturation, and it, not a count, is the real finish line.
A narrative review (this workflow, used flexibly) selects and interprets literature to build an argument. A systematic review pre-registers a protocol, searches exhaustively against it, and reports every decision so another researcher could reproduce the result. Systematic reviews are the standard for clinical and policy questions; narrative reviews suit theory-building and dissertation chapters.
For discovery help — suggesting synonyms, mapping a field's vocabulary, summarising a paper you then read properly — most universities allow it. The screening decisions, the reading and every word of the writing must be yours, and every source must be one you have actually opened. Disclose tool use if your institution asks.
You do not, quite — the steps overlap. But the signal to shift your centre of gravity to writing is saturation: the same names and findings keep recurring, and new papers slot into existing themes instead of creating new ones.
Keep reading
Your thesis topic will live with you for months, so choose it properly. Learn how to brainstorm widely, test ideas against interest, feasibility and contribution, and narrow a vague interest into a question you can actually finish.
A research proposal is a promise and a plan. Learn the sections reviewers expect, how to frame a sharp question, write a methodology that survives scrutiny, and the feedback loop that turns a draft into an approval.
Research papers are not novels. Learn the three-pass method — skim, read, interrogate — with time budgets, a comparison table and the five-question note that makes papers stick.