Skip to content
Analytics

Crawlers, citations and referrals are three different signals

Keep automated requests, answer observations and human visits separate when reviewing AI search performance.

Obility Editorial · · 3 min read

A request is not an endorsement

A server request can tell you that a URL was requested and what user agent was supplied. It does not establish that an answer engine recommended the page or that a person read it.

When classifying bots by user agent, label that method clearly. A user-agent string can be imitated; it is not the same as independently verified crawler identity.

A citation belongs to a particular answer

A citation observation should include the question, answer, engine and collection context. Without that evidence, a citation count is difficult to interpret or reproduce.

Keep track of the destination URL and why it appears relevant. Distinguish a source supporting one sentence from a recommendation of the company behind that source.

A referral is a visit to your site

Website analytics can help you understand visitors who arrive with an identifiable referral. Review the landing page and the actions that follow, such as reading an article or requesting a demo.

Not every visit will carry enough referral information to attribute it confidently. Avoid assigning all unexplained direct traffic to AI search. Keep uncertainty visible in the report.

Design a report that preserves the distinctions

Use separate views for crawler requests, observed citations and referred visits. Define each metric near the chart. Compare trends only when their scope and collection methods are compatible.

Connect the report to decisions: a crawler issue may call for a technical check; an inaccurate answer may call for a fact review; a weak landing-page conversion may call for a clearer page. Each signal should guide an appropriate next step.

Put the workflow into practice.

Bring your questions and content to an Obility demo.

Book a demo