From fifth to second: One Identity’s external-citation approach to LLM optimization

You improve your own website, but AI keeps citing outdated comparison pages. In that situation, it is useful to investigate the external sources supporting the answers too.

One Identity reportedly improved its average competitive position from approximately fifth to second. Its work included reviewing cited external articles and engaging their publishers. Profound’s case study

The case considers LLM optimization beyond a company’s own site.

What One Identity’s move from fifth to second and 30% visibility increase mean

The move from approximately fifth to second concerns average positioning against the selected competitors. The story also reports a 30% visibility increase in one quarter. Profound’s case study

Illustration: Distinguish competitive positioning in AI from placement inside an external article.

This is not Google search ranking. It is also separate from first- or second-position placements within publishers’ comparison articles.

These vendor-reported figures do not demonstrate higher revenue or more contracts. Start by identifying which questions and sources lead to the recommendations.

Using 90 days of citations to select external comparison pages for outreach

One Identity used a 90-day view of citations to identify external comparison pages for publisher outreach. Sponsored placements were part of the activity. Profound’s case study

Illustration: Review recurring external citations over a defined period and distinguish sponsored placements.

An appearance in an external article is not automatically an independent endorsement. Distinguish factual corrections, editorial coverage, and paid placement.

For your own company, first look for outdated prices or incorrect specifications. An official source can support a concrete correction request.

A vendor-authored Active Directory comparison also supported the work

The company also created an Active Directory management tools comparison on its own site. Profound’s case study

The current One Identity comparison describes multiple tools and places its own Active Roles product first. It is a vendor-authored comparison, not an independent ranking.

The current page may differ from the version used during the case period. Review its decision criteria rather than imitating its ordering.

Build a time-bounded list of external pages cited for your questions

Create a list of external pages cited in answers to a stable question set. This is our proposed workflow.

Illustration: Open cited pages and compare their claims with current, verified company facts.

Field	What to record
Question and date	Conditions under which the answer appeared
Cited URL	The article, not just the domain
Description of your company	Capabilities, audience, or pricing
Difference from official evidence	Outdated, incorrect, or missing information
Placement type	Editorial, advertising, or unknown

A fixed window such as 90 days helps distinguish recurring citations from one-off observations. It does not mean waiting 90 days before requesting a factual correction.

When contacting a publisher, identify the exact passage and official evidence. Keep buying a placement separate from correcting inaccurate information.

Check paid-placement disclosure and track AI position separately from visits

For paid placements, ask the publisher about advertising disclosure and link treatment. Google recommends rel="sponsored" for paid links and also accepts nofollow. Google’s link-attribute guidance

Illustration: Identify advertising and qualify paid links while tracking AI positioning and visits separately.

Those attributes do not promise increased AI visibility. Evaluate advertising placement and changes in AI answers separately.

Record the external page’s update date, average AI position, citations, and visits to your site independently. A prominent sponsored listing should not become a claim of a higher position across AI systems.

Start by reading three external pages cited for your main product and checking whether their facts are correct.