Prompt · Research

Literature review matrix from your sources

Organise the papers or reports you have collected into a comparison matrix of questions, methods, samples and findings, then see where studies agree, conflict and leave gaps.

The prompt

You are a research assistant experienced in systematic reading. Build a literature review matrix from the sources I provide. Work only from the text I paste.

My research question: [QUESTION]
Field: [e.g. "public health", "education", "HCI"]
What I need the review for: [e.g. "thesis chapter", "grant proposal background", "internal report"]

Sources (each labelled, with abstract, methods and results sections where possible):
"""
[P1] [AUTHORS, YEAR, TITLE] [PASTED TEXT]
[P2] ...
"""

Step 1 - Matrix. A table with columns:
ID | Authors & year | Research question | Method / design | Sample & setting | Key findings | Limitations stated by the authors | Relevance to my question (high/medium/low)
Write "NOT REPORTED" when the pasted text does not include the information.

Step 2 - Synthesis:
- Themes: where do 2 or more sources agree? Cite IDs.
- Conflicts: where do findings disagree, and could differences in method or sample explain it?
- Gaps: what my question needs that no source addresses.
- Quality notes: which findings rest on small samples, short time frames or self-reported data.

Rules: do not add studies, authors or statistics I did not provide. Quote exact numbers only as they appear in the text. If a claim in the synthesis is your interpretation, label it "INTERPRETATION".

Example output

Example (excerpt):
| ID | Authors & year | Method / design | Sample & setting | Key findings | Relevance |
| P1 | Rossi et al., 2023 | Randomised trial, 8 weeks | 212 secondary students, Italy | Retrieval practice group scored 11 points higher on the final test | high |
| P2 | Chen, 2021 | Survey | 1,050 university students, online | Most students reported re-reading as their main strategy | medium |

Conflict: P1 and P4 disagree on effect size. INTERPRETATION: P4 measured after 1 day, P1 after 8 weeks.
Gap: no source studies adult learners outside formal education.

When to use it

Use this prompt once you have gathered a set of sources and need to see them side by side: for a thesis chapter, a grant application, a policy brief or a background report at work. It is most useful with five to fifteen sources, when the details start to blur and it becomes hard to remember which study used which method.

How to customize it

[RESEARCH QUESTION] is what makes the “Relevance” column and the gaps meaningful. A sharper question gives a sharper review.
[FIELD] sets expectations about methods and terminology.
[WHAT I NEED THE REVIEW FOR] changes the depth. A grant proposal needs gaps; a thesis needs careful method comparison.
[SOURCES] should include the methods and results, not just the abstract. Abstracts often leave out limitations and sample details. For many sources, process them in batches and merge the tables.

Why it works

A matrix is a long-standing tool for literature reviews because it forces the same questions for every source. Patterns, conflicts and gaps become visible when you compare rows, rather than reading summaries one after another.

The prompt keeps the model inside your material. Language models can produce realistic-looking citations for studies that do not exist, a well-documented form of hallucination. By pasting the text and banning added sources, you limit the work to extraction and comparison, which models do far more reliably. “NOT REPORTED” makes missing information visible instead of guessed, and the “INTERPRETATION” label keeps reasoning separate from what the authors actually wrote.

Tips and variations

Keep a column for the full reference in your preferred citation style, filled in by you from the original paper rather than generated.
If your assistant can search the web or academic databases, ask it to suggest more sources in a separate step, then find and read each one yourself before adding it.
Export the matrix to a spreadsheet so you can sort by method, year or relevance.
Check every number in the table against the original paper before you cite it.
Our guide on how to read an AI research paper has a useful reading routine for any field.
For a single-question overview, see the research brief with source checking prompt.

#Literature Review #Research #Academic Writing #Synthesis

More Research prompts

View all

Researchintermediate

Fact-check list for every claim in a draft

You are a fact-checker at a careful publication. Do not judge whether claims are true from memory. Your job is to build a verification checklist. Type of text: [e.g. "blog post", "company report", "speech", "social media thread"] Audience and stakes: [e.g. "public article, legal risk if wrong"] Sources the author used (if known): [LIST, or "unknown"] Steps: 1. Extract every checkable claim: numbers, dates, names and titles, quotes, causal statements ("X led to Y"), comparisons ("the largest", "the first") and descriptions of studies or laws. 2. For each claim give, in a table: # | Claim (exact words) | Type | Risk if wrong (high/medium/low) | What would verify it (primary source to look for) | Notes 3. Mark as high risk: statistics, quotes, health, legal or financial statements, and anything about a named person or company. 4. Flag vague claims that cannot be checked as written ("experts say", "studies show") and suggest how to make them specific. 5. List claims where the wording goes beyond what a typical source would support (e.g. "proves" instead of "suggests"). If you can browse the web: for high-risk claims only, look for a primary source, give the link and quote the relevant sentence. Mark each as "SUPPORTED", "CONTRADICTED" or "NOT FOUND". If you cannot browse, say so and do not assign these labels. Draft: """ [PASTE TEXT] """

Works with: ChatGPT, Claude, Gemini, Perplexity

Researchintermediate

Research brief with source checking

You are a careful research analyst. Prepare a research brief on this question: [RESEARCH QUESTION] Purpose of the brief: [WHAT DECISION OR PIECE OF WORK IT SUPPORTS] Time frame that matters: [e.g. "studies since 2022"] Sources to use: [PASTE SOURCES, OR WRITE "search the web" IF YOUR ASSISTANT CAN BROWSE] Work in this order: 1. Restate the question and list 3-5 sub-questions that together answer it. 2. For each sub-question, give the key findings. After every factual claim, cite the source in brackets. 3. Build a source table: Source | Type (peer-reviewed, official, news, opinion, vendor) | Date | Reliability (high/medium/low) | Why. 4. List points where sources disagree and explain the likely reason. 5. List what you could not find or verify. Do not fill gaps with assumptions. 6. End with a 5-sentence answer to the original question and a confidence level (high/medium/low). Rules: - Use only the sources provided or found. If you cannot access a source, say so. - Never invent titles, authors, URLs or statistics. - Mark any claim that relies on a single source as "single source".

Works with: ChatGPT, Claude, Gemini, Perplexity

Researchintermediate

Survey question design and bias review

You are a survey methodologist. Help me design a survey that produces reliable answers. What I want to learn: [2-4 RESEARCH GOALS] Respondents: [WHO, AND HOW THEY WILL RECEIVE IT, e.g. "customers by email, mostly on mobile"] Maximum length: [e.g. "5 minutes", "12 questions"] Decisions the results will inform: [e.g. "which feature to build next"] Existing draft (optional): """ [PASTE DRAFT QUESTIONS, or "none"] """ If I gave a draft, review each question first: - Flag problems: leading or loaded wording, two questions in one, vague terms ("often", "regularly"), jargon, overlapping or missing answer options, questions respondents cannot answer accurately. - Give an improved version and one line explaining the change. Then produce the final survey: 1. A one-sentence introduction that states purpose and time needed. 2. Questions in a logical order: easy and general first, sensitive or demographic questions last. 3. For each question: the type (single choice, multiple choice, rating scale, open text), the answer options, and which research goal it serves. 4. Use balanced rating scales with labelled points, and include "Not applicable" or "Prefer not to say" where needed. Finish with: questions I could remove if the survey is too long, and a suggestion to test it with 3-5 people first.

Works with: ChatGPT, Claude, Gemini, Any LLM