CitabilityFree

Original Research Signals

Detects phrases like 'our research', 'we found', 'survey of', and custom charts. Original research gets cited far more than derivative content.

Why It Matters for AI Visibility

AI engines like ChatGPT and Perplexity strongly prefer original primary sources over content that rehashes existing findings. When multiple pages cover the same topic, the one with original research, survey data, or case studies gets cited as the authoritative source. Derivative content gets skipped. Google AI Overviews prioritize unique data that cannot be found elsewhere. Original research is inherently unique -- no one else has your survey results, your case study outcomes, or your proprietary analysis. This makes your page the only possible citation for that specific data point. The practical difference is stark. A blog post saying "studies show that remote work increases productivity" competes with thousands of identical statements. A post saying "we surveyed 500 remote workers and found that 73% reported higher productivity" is a primary source that AI engines will reference directly. First-party data, documented methodology, and visual data presentations all signal to AI crawlers that your content contributes new knowledge rather than repeating existing information.

How We Score It

The analyzer detects two signal types: research language phrases and chart or graph HTML elements. Research phrases include first-person language like "we found," "our research," "we surveyed," and methodology terms like "we benchmarked" or "we interviewed." Charts are detected through HTML elements such as Canvas (Chart.js), SVG with chart-related classes, and elements from libraries like D3, Highcharts, Recharts, or Plotly. Zero research phrases and no charts scores 0. A single phrase without charts scores 5. Two phrases reach 7. Three or more phrases without charts, or any phrases combined with chart elements, scores 8. The maximum score of 10 requires either 3+ phrases with charts present or 5+ unique research phrases. Each unique phrase is counted once after deduplication.
See how your site scores on this factorAnalyze My Site

How to Improve

  • 1

    Use first-person research language

    Write "we analyzed 1,000 websites" instead of "an analysis of websites shows." The analyzer specifically detects phrases like "we found," "our research," "our data shows," and "we surveyed." Each unique phrase counts toward your score, so vary your research language across the page.

  • 2

    Add a case study section

    The phrase "case study" is a detected signal. Document a real customer outcome, an internal experiment, or a before-and-after comparison with specific numbers. A single case study section adds both a research phrase and substantive original content that AI engines can cite.

  • 3

    Include data visualizations using chart libraries

    The analyzer detects chart HTML elements from Chart.js, D3, Highcharts, Recharts, and Plotly. Adding an interactive chart alongside your research data contributes significantly to the score. Static images of charts do not count -- you need actual chart library elements in the HTML.

  • 4

    Describe your methodology explicitly

    Phrases like "we interviewed 50 customers," "we benchmarked 200 sites," and "we tracked results over 6 months" each count as unique research signals. Describing how you gathered data strengthens both your score and the credibility AI engines assign to your findings.

  • 5

    Label your research as original

    Use ownership signals like "our proprietary data," "original research," and "our original analysis." These phrases are explicitly detected by the analyzer and help distinguish your content from pages that merely reference others' work.

Before & After

Before
Studies show that content marketing generates 3x more leads than
traditional marketing. According to industry reports, companies
that blog get 67% more leads.
After
We analyzed 500 B2B companies over 12 months and found that content
marketing generated 3.2x more leads. Our data shows that companies
publishing weekly saw 67% more leads. We surveyed 200 marketing
directors to confirm these findings.

[Chart.js bar graph showing lead generation by publishing frequency]

Code Examples

Chart elements detected by the analyzer

<!-- Chart.js canvas (detected) -->
<canvas id="research-chart" width="600" height="400"></canvas>

<!-- D3 visualization (detected by class pattern) -->
<div class="d3-chart" id="findings-visualization"></div>

<!-- Recharts container (detected by class pattern) -->
<div class="recharts-wrapper"></div>

Frequently Asked Questions

Does citing others' research count toward this score?

No. This factor specifically detects first-person research language like "we found" and "our data." Citing external studies is evaluated by different factors (External Citations and Source Attribution). This factor rewards content where you are the primary source.

Can I improve my score by adding research phrases without actual research?

The analyzer detects phrase patterns, so the phrases would technically count. However, AI engines evaluate actual content quality when deciding what to cite. Genuine research with supporting data will always outperform keyword-stuffed content. The phrases are meant to reflect real research, not game the score.

Do embedded images of charts count, or must they be interactive chart elements?

Only HTML chart elements are detected -- Canvas elements, SVGs with chart-related classes, and elements from known chart libraries (Chart.js, D3, Highcharts, Plotly, Recharts). Static images of charts in `<img>` tags are not detected. You need actual chart library elements rendered in the HTML.

Related Factors

Check Your GEO Score

Run a free analysis on your website and see how you score across all 52 factors.

Analyze My Site