Almost every rejection letter that flags a “descriptive” literature review is pointing at the same problem: the author is summarising, not synthesising. The paragraphs read like sequential book reports — “Smith found this. Jones found that. Lee found something else.” — and reviewers conclude the author has read the field without thinking about it.
This guide shows you exactly how to synthesise a literature review, with worked before-and-after examples. The good news is that the difference is not about reading more — it is about rewriting what you have read using a small set of repeatable moves.
A summary tells the reader what each individual paper says. A synthesis tells the reader what a group of papers, taken together, reveals about the field. The unit of analysis shifts from the paper to the argument.
The mechanical sign of a synthesis paragraph is simple: each paragraph opens with a claim about the field, not with an author name and a year. Citations then appear as evidence for that claim, often clustered together: (Smith, 2018; Jones, 2019; Lee, 2020).
This is so consistent that you can audit your own literature review in 60 seconds. If every paragraph starts with “Author (Year) found…”, you are summarising. If paragraphs start with statements about the field, you are synthesising.
Below are three before-and-after pairs from real-feeling research scenarios — management, education, and engineering. Each pair includes the rewrite moves used, so you can apply the same techniques to your own paragraphs.
Smith (2018) found that transformational leadership improves engagement. Jones (2019) found that transactional leadership improves engagement. Lee (2020) found that authentic leadership improves engagement. Patel (2021) found that servant leadership improves engagement.
Despite four competing leadership styles being independently linked to engagement (Smith, 2018; Jones, 2019; Lee, 2020; Patel, 2021), none of these studies has tested the underlying mechanism. The literature confirms an effect across multiple leadership constructs, but leaves unresolved the process through which leadership translates into engagement — particularly in hybrid work contexts.
Chen (2023) studied generative AI use in higher education classrooms. Wang (2024) studied generative AI use in higher education classrooms. Kumar and Mehta (2024) studied generative AI use in higher education classrooms in India. Tan and Lim (2025) studied generative AI in Singaporean universities.
Empirical work on generative AI in higher education has expanded rapidly since 2023 (Chen, 2023; Wang, 2024), but most studies remain confined to elite or well-resourced institutional contexts (Kumar & Mehta, 2024; Tan & Lim, 2025). The lived experience of faculty in resource-constrained, non-elite institutions — where AI adoption decisions are entangled with workload pressures — remains largely unexamined.
Several deep learning models have been proposed for predictive maintenance. Author A used CNN. Author B used LSTM. Author C used a hybrid CNN-LSTM. Author D used a transformer-based model.
Recent deep learning approaches to predictive maintenance have evolved from single-architecture models (CNN, LSTM) toward hybrid and attention-based designs (Author C; Author D), each claiming improved accuracy. However, comparisons across studies are confounded by the use of different benchmark datasets and inconsistent evaluation metrics, making a clean assessment of which architecture genuinely generalises across industrial contexts difficult.
Three different fields, three different framings — but the same underlying technique. Whatever your discipline, knowing how to synthesise a literature review comes down to a small toolkit of moves you can apply paragraph by paragraph.
If you remember nothing else from this guide, remember these five moves. Each one transforms a summary into a synthesis paragraph.
Lead the paragraph with a statement about what the field, taken together, has shown — not with a single author.
Group three or more sources that share a finding into a single parenthetical. Visual density signals depth of reading.
Tell the reader what the grouped citations reveal — agreement, disagreement, evolution, or a shared blind spot.
Most synthesis gaps live here. What do these studies, collectively, fail to address?
Point forward to what remains unknown. This sets up either the next paragraph or — at the section level — your research gap.
You will not use all five in every paragraph. Most strong synthesis paragraphs use three or four. The combination matters more than the count.
Open your draft and count how many distinct citations appear in each paragraph. In strong literature reviews, most paragraphs contain three or more citations, often clustered. Paragraphs with one citation typically read as descriptive — even when the writing itself is competent.
If most of your paragraphs cite only one source, your review is probably summarising. Reorganising those paragraphs so that related studies are discussed together is the single most effective revision you can make.
When stuck, list out the three to five key arguments the field has made on your topic. Drag each cited paper under the argument it best supports. Now write one paragraph per argument, citing all the supporting papers together. This single reorganisation turns a summary into a synthesis.
A paragraph that begins with an author's name and ends with their finding has done the work of a footnote, not a literature review.
A 20-minute consultation can identify which paragraphs need restructuring — and how.
Synthesis is sometimes confused with two other moves that look similar but do not satisfy reviewers.
Synthesis is not paraphrasing. Rewriting a single source in your own words is still summarising. Synthesis requires combining multiple sources into a higher-order claim.
Synthesis is not opinion. Adding your view between citations — “This finding is interesting…” — does not turn a summary into a synthesis. Synthesis is what the literature, taken together, shows. Your interpretation comes later, in the discussion.
Synthesis is not citation-stuffing. Putting eight references at the end of one sentence is not synthesis if you have not actually engaged with what they have in common. Reviewers can tell. A clean cluster of three carefully chosen sources beats eight padded ones.
Author-per-paragraph structure. One paragraph for Smith, then one for Jones, then one for Lee. Reorganise by argument instead.
The “laundry list” opener. “Many studies have explored X (Smith, 2018; Jones, 2019; Lee, 2020).” reads as a citation dump. Tell the reader what those studies, collectively, found — not just that they exist.
Failure to point forward. Synthesis paragraphs should end with a tension or an unresolved question, not with a definitional statement.
Synthesis only at section start, summary everywhere else. Reviewers read full sections. A strong opening paragraph cannot rescue four paragraphs of summary that follow.
Pick the longest section of your current literature review. Count how many citations appear in each paragraph. Find the paragraph with the most one-source paragraphs and rewrite those first — reorganise by argument, cluster citations, and add the unresolved question at the end. Then repeat for the rest of the section.
Knowing how to synthesise a literature review is one of the highest-leverage skills in academic writing. Get it working in one paper and every future paper benefits. For the full structural picture, see our pillar on how to write a literature review that positions your research gap, and our companion piece on thematic vs chronological vs methodological literature reviews.
If you would like a second pair of eyes flagging the paragraphs that need synthesis work, that is exactly what the literature review audit inside our manuscript preparation service covers.
Book a free consultation. We will read your literature review and flag the paragraphs that need restructuring — with specific synthesis moves to use.
Book Free Consultation Check via the Health Checker