Understanding sources and citations
Source analysis examines the websites and pages used in AI answers. Find information supporting your brand, articles featuring competitors, and repeatedly cited comparisons to guide content improvements.
Why investigate sources?
You control the content on your own website. For external sites, information sharing, partnerships, and community participation can provide opportunities to correct outdated information or fill gaps. You cannot prescribe the AI's answer, but you can improve the accuracy and completeness of the information it may reference.
Use domains and URLs together
For example, find a publication frequently associated with competitors, open its comparison article, then read AI answers to see which statements cite it.
Match the approach to the source type
| Source type | Example approach |
|---|---|
| Editorial | Provide primary information to journalists or editors and check factual accuracy |
| Corporate | Supply relevant information to partners or industry directories |
| UGC | Answer questions and share useful information within community rules |
| Reference | Use the established editing or correction process to fill gaps |
| Your website | Organize headings, definitions, comparisons, and evidence for readability |
At URL level, inspect formats such as articles, listicles, how-to guides, and comparisons. Use the formats prevalent in your industry to understand how to answer the relevant questions.
Retrievable content varies by model and collection method. Check whether important information is behind login, a paywall, or JavaScript execution, and verify that it appears in the collected page text.
The Alpha POST /v1/reports/social-citations operation ranks saved post URLs
from YouTube, Reddit, LinkedIn and Instagram. It reads citation facts; it does
not collect new social content or infer authors or channels. Profile and home
pages are excluded. Each row includes the exact observed post_url, socialplatform, citation_count, distinct answer_count, rank and citation share
as a 0–1 ratio. Different observed URLs remain separate ranking keys.
Supply inclusive UTC start_date and end_date. platforms selects one or
more of youtube, reddit, linkedin, instagram; prompt_ids optionally
limits citations to saved prompts. Omit prompt IDs or send [] for all prompts.
Shares and totals cover all matching posts before limit/offset pagination.
Unknown platforms, dimensions and filters return 400. All reads are workspace scoped.
orc reports social-citations --start-date 2026-11-01 --end-date 2026-11-02 --platforms youtube,reddit,linkedin,instagramMCP: get_social_citation_report accepts the same JSON fields.
Recognized post paths are YouTube /watch, /shorts/ and /live/ (also youtu.be); Reddit /comments/, /r/{community}/comments/ and /gallery/; LinkedIn /posts/, /feed/update/ and /pulse/; and Instagram /p/, /reel/, /reels/ and /tv/.