← Back to Tutorials
Research intermediate Elena Torres ·

Using Scite and Consensus for Academic Literature Review 2026

Using Scite and Consensus for Academic Literature Review 2026

Overview

A thorough literature review is the foundation of good research — but it’s also one of the most time-consuming parts of academic work. Finding relevant papers, assessing their credibility, understanding how they’ve been cited, and synthesizing findings can take weeks.

Scite and Consensus are two AI-powered research tools that dramatically accelerate this process. Together, they form a powerful literature review workflow:

  • Scite shows you how papers are actually used by other researchers — not just citation counts, but whether citations are supporting, contrasting, or mentioning
  • Consensus is a search engine for academic papers that uses AI to extract findings, evaluate study quality, and provide direct answers to research questions

This tutorial covers a complete literature review workflow using both tools — from searching and filtering, to extracting insights, to generating a first-draft literature review.

Who this is for: Graduate students, academic researchers, PhD candidates, and anyone conducting structured literature reviews.

Prerequisites

  • Scite account (free tier available at scite.ai; Pro recommended for serious research)
  • Consensus account (free tier at consensus.app; Premium for unlimited searches)
  • Reference manager (Zotero free, or Mendeley, EndNote) for saving papers
  • A research question or topic you want to review

Step-by-Step Guide

Step 1: Set Up Your Scite Account and Dashboard

  1. Go to scite.ai and sign up (email or Google SSO)
  2. The free tier includes:
    • 50 “See How It’s Cited” lookups per month
    • 25 “Search” queries per month
    • Basic citation statement viewing
  3. For this tutorial, the free tier is sufficient — but if you’re doing a full literature review, the Pro plan ($20/month academic pricing) is worth it

Configure your dashboard:

  1. Click your profile → Settings → Dashboard
  2. Add widgets for:
    • Recent searches
    • Saved paper lists
    • Citation statement feed (papers citing your saved papers)
  3. Set your Research Discipline — this helps Scite prioritize relevant sources

Install the browser extension: The Scite browser extension (Chrome/Firefox) adds a sidebar to any paper page showing how that paper has been cited. Install it — it’s invaluable during exploratory reading.

  1. Go to scite.ai → Tools → Browser Extension
  2. Install for your browser
  3. When you visit a paper on PubMed, arXiv, Google Scholar, or a publisher site, click the Scite icon
  4. See citation statements and classification immediately

Standard citation counts tell you nothing about how a paper is being used. A paper could be widely cited as “this methodology is flawed” and still have high citation numbers. Scite’s Smart Citations solve this by classifying each citation:

  • Supporting — The citing paper uses this as evidence or agrees with the finding
  • Contrasting — The citing paper disagrees, finds contradictory evidence, or notes limitations
  • Mentioning — Neutral mention, no evaluative stance

Perform a smart citation search:

  1. Click “Search” in Scite’s top bar
  2. Enter a paper title or DOI:
Microdosing psychedelics: personality, mental health, and creativity differences in microdosers

  1. Scite shows:

    • Total citation count (e.g., 187 citations)
    • Supporting vs Contrasting breakdown (e.g., 45% support, 12% contrast, 43% mention)
    • Citation statement previews — actual sentences from citing papers

  2. Click “View Citations” to see the full breakdown with detailed statements

Filter citations by:

  • Year range — Are recent citations different from older ones?
  • Journal quality — Filter by high-impact vs. lower-tier journals
  • Discipline — How is this paper used in psychology vs. neuroscience?
  • Support/Contrast — Focus specifically on papers that challenge the finding

Practical example: Search for “Does ketamine therapy work for depression?” in Consensus first (we’ll cover this next), then go to Scite to check how the key papers are being cited. If a landmark study has 80% supporting citations, the evidence base is strong. If it has 30% contrasting citations, the findings are contested — and that’s important context for your review.

Step 3: Use Scite to Evaluate Individual Papers

When you’ve found a paper you’re considering including in your review, Scite helps you evaluate it:

  1. Enter the paper’s DOI or URL in Scite’s search
  2. Look at the Citation Distribution Graph — it shows citations over time, colored by support/contrast
  3. Check “Cited by top journals” — are high-impact journals citing this paper?
  4. Review “Contrasting citations” first — these tell you the paper’s limitations and criticisms
  5. Review “Supporting citations” — what specific findings are being backed up?

Red flags to watch for:

  • High contrast rate (>20%) — The paper’s conclusions are contested
  • Clustering of contrasting citations in recent years — Newer evidence may be overturning older findings
  • Citations only from low-impact journals — The paper may not have penetrated the main discourse
  • No citations from your target journals — If your review is for a top-tier journal, the cited papers should include those sources

Green flags:

  • High support ratio (>60%) — Widely accepted findings
  • Citations from diverse disciplines — Broad impact
  • Increasing citation rate — Growing relevance

Step 4: Set Up Consensus for Answer-Based Research

Consensus is a fundamentally different tool from Google Scholar. Instead of returning a list of paper titles, it answers your research question directly with evidence from papers.

  1. Go to consensus.app and sign up
  2. The free tier gives you 20 searches/month
  3. Install the Consensus browser extension as well — it adds a sidebar to PubMed and Google Scholar

Write effective research questions for Consensus:

Too VagueBetter
”What is the effect of exercise?""Does high-intensity interval training improve cardiovascular health in adults over 50?"
"Cannabis and memory""Does regular cannabis use affect short-term memory performance in young adults?"
"AI in medicine""What is the accuracy of AI-powered diagnostic tools for detecting skin cancer?”

The difference: Consensus works best with a testable hypothesis phrased as a question. It finds papers that directly address that question.

Run your first Consensus search:

  1. Enter your research question:
Does intermittent fasting lead to greater weight loss compared to daily calorie restriction?
  1. Consensus returns:
  • Top papers with study summaries generated by AI
  • Consensus meter — An overall finding (e.g., “Intermittent fasting may be more effective for weight loss based on 42 studies”)
  • Study characteristics — Sample size, population, study type (RCT, meta-analysis, observational)
  • Extracted findings — One-line summaries of each paper’s conclusion
  1. Click any paper card to see:
    • Full abstract
    • Sample size and population
    • Study type (RCT, systematic review, etc.)
    • Link to the full text

Step 5: Extract Insights from Consensus Results

Consensus’s AI extraction is its killer feature. Each search returns structured data you can copy directly into your literature review notes:

Export findings:

  1. After searching, click “Export” (top right of results)
  2. Choose format:
    • CSV — For spreadsheet analysis
    • BibTeX — Import into Zotero or Mendeley
    • Plain text — For pasting into notes

The CSV export includes columns: Title, Authors, Year, Journal, Study Type, Sample Size, Finding Summary, URL. This is your literature review table ready-made.

Use Consensus’s Study Type filter: Filter results to show only:

  • Meta-analyses — Highest level of evidence
  • Systematic reviews — Comprehensive literature summaries
  • Randomized controlled trials — Gold standard for causal claims
  • Observational studies — Real-world evidence

Pro tip: Run your Consensus search first, then export to CSV. This gives you a structured list of papers. Then take each paper to Scite for detailed citation analysis.

Step 6: Cross-Reference with Scite for Credibility

Once Consensus gives you a list of relevant papers, use Scite to verify each one:

Workflow: Consensus → Scite cross-reference

  1. From Consensus, note the DOI or title of the strongest paper
  2. Paste it into Scite’s search
  3. Check the support/contrast ratio
  4. Read 3-5 contrasting citation statements to understand critiques
  5. Read 3-5 supporting citation statements to see corroboration
  6. Make a judgment: Is this paper reliable enough to include?

Example cross-reference: From Consensus: “Intermittent fasting shows significant weight loss compared to calorie restriction (RR 1.15, 95% CI 1.02-1.30)” based on a 2024 meta-analysis.

Scite check on that meta-analysis:

  • 67 citations: 52% supporting, 8% contrasting, 40% mention
  • Supporting citations: Clinical trials confirming the finding
  • Contrasting citations: 2 papers suggesting the effect disappears at 12-month follow-up

This critical nuance — that short-term benefits may not persist — is exactly the kind of insight that elevates a literature review from “list of papers” to “synthesis of evidence.”

Step 7: Save and Organize with Your Reference Manager

Both Scite and Consensus integrate with reference managers:

From Scite:

  1. Click “Save” on any paper
  2. Export options: BibTeX, RIS, Zotero direct
  3. Create Collections in Scite — organize papers by theme

From Consensus:

  1. Click the bookmark icon on any paper
  2. Create Lists organized by research question
  3. Export your list as BibTeX for Zotero import

Zotero integration workflow:

  1. Install the Zotero browser connector
  2. When viewing a paper in Consensus or Scite, click the Zotero icon
  3. Zotero captures metadata automatically
  4. Tag papers in Zotero: #scite-reviewed, #consensus-confirmed, #contrasting-evidence

This ensures your references are organized before you start writing.

Step 8: Generate Your Literature Review Draft

With both tools, you can generate a first-draft literature review:

Using Consensus’s AI synthesis:

  1. Run a comprehensive search on your research question
  2. Click “Synthesize” (Pro feature) to get a narrative summary
  3. The synthesis covers: what’s known, points of disagreement, gaps in evidence, and directions for future research
  4. Copy this as your starting draft

Using Scite’s Assistant:

  1. Build a collection of 10-20 key papers in Scite
  2. Use Scite Assistant (available in Pro):
    • “Summarize the key findings from this collection”
    • “What are the main points of disagreement between these papers?”
    • “Identify research gaps across these papers”
  3. The Assistant’s output follows academic conventions — use it as a literature review framework

Bringing it together — your writing workflow:

1. Define research question
2. Search Consensus → get structured list of papers
3. Filter by study type (prioritize meta-analyses and RCTs)
4. Export to CSV → get structured data table
5. For each key paper: Scite citation analysis
6. Note contrasting citations → these become your "debate" section
7. Use Consensus synthesis + Scite Assistant → generate draft
8. Cross-reference drafts with Zotero library
9. Add your own analysis and synthesis
10. Write final literature review

Expected time savings: Without AI tools: 2-3 weeks for a comprehensive literature review With Scite + Consensus: 2-4 days for the first draft

Troubleshooting

Scite says “no citation data found”

Some papers, especially recent preprints, haven’t been cited enough for meaningful analysis. Use Scite’s “Search by Topic” instead — enter keywords and Scite will find all relevant papers with citation data.

Consensus returns irrelevant results

Your research question may be too broad or poorly phrased. Refine it:

  • Add population (“in adults over 65”)
  • Add intervention (“compared to placebo”)
  • Add outcome (“measured by blood pressure”)
  • Add time frame (“over 12 weeks”)

Too many papers to review

Use Consensus’s Study Type and Year filters aggressively. For a focused review, limit to the last 5 years and meta-analyses/systematic reviews only. You can always expand if needed.

Cannot access full text

Use the Unpaywall or Open Access Button browser extension alongside Scite. Many papers Scite indexes have open-access preprints. Alternatively, check Google Scholar or your institution’s library portal.

Next Steps / Advanced

  1. Scite Reference Check — Before submitting your paper, run Scite Reference Check (Pro feature). It analyzes your entire reference list, flags retracted papers, and shows how each reference has been cited.

  2. Consensus API — Consensus offers an API for programmatic search. Build an automated literature screener using Python:

import requests

# Find papers on your research question
url = "https://api.consensus.app/search"
params = {
    "q": "Does meditation reduce anxiety in clinical populations?",
    "limit": 20,
    "sort": "citation_count"
}
response = requests.get(url, headers={"Authorization": "Bearer YOUR_KEY"})
papers = response.json()

  1. Scite + n8n integration — Set up an n8n workflow that monitors Scite for new citations to your saved papers. Get a notification when important new papers cite your key references.

  2. Create a living literature review — Use Scite Collections and Consensus GPT (available in ChatGPT’s plugin store) to maintain an up-to-date literature review that automatically adds new papers as they’re published.

FAQ

Are Scite and Consensus free?

Both have free tiers with limited usage. Scite free: 50 citation lookups/month. Consensus free: 20 searches/month. For a full literature review, expect to need at least one Pro subscription ($20/month academic pricing).

How accurate is Consensus’s AI summary?

Consensus claims 95%+ accuracy in extracting findings from abstracts. However, always verify the AI summary against the actual paper — especially for conclusions that seem surprising or contradictory to your existing knowledge.

Can I use these tools for non-academic research?

Absolutely. Both tools work well for policy research, market research reports, and evidence-based business analysis. The literature review workflow is the same regardless of domain.

What about Google Scholar and PubMed?

Neither Scite nor Consensus replaces Google Scholar and PubMed — they augment them. Use Google Scholar for initial exploration and broad searches, Consensus for targeted answer-based research, and Scite for credibility assessment. The best workflow uses all three.