Larridin is a leader in measuring your company’s use of AI and optimizing AI’s contribution to business results. One of those business results is how your brand appears in LLM search results.
A new CMSWire piece from Lawrence Shaw is the third in his series on AI readiness. It lays out just how wrong AI can be in results about your brand. get it, why it happens, and what digital teams can do. And it mentions the importance of AI measurement and optimization as part of AI readiness.
How AI Confidently Misrepresents Brands
The results are worse than most leaders assume:
- A national financial regulator published key legal documents as PDFs. Shaw found that 84% of AI-generated summaries of those documents contained inaccuracies, including errors about the core legal basis for regulation.
- An AI chatbot on a local government website served citizens incorrect planning information because it treated five-year-old data from an orphaned departmental site as the authoritative source.
And most of what AI says about you isn't even coming from you. McKinsey estimates that up to 90% of what AI search references sits on third-party sites outside a brand's control; AirOps puts third-party brand mentions at 85%. Meanwhile, only 14% of marketers track AI visibility at all, so they literally don’t see the problem. .
Nobody Owns the Problem
How does this happen? Three big reasons:
- Unclear ownership. Larridin research shows that 58.2% of organizations say "unclear and fragmented ownership" is the main barrier to measuring AI performance, and 75% lack AI governance entirely.
- Digital estate sprawl. Shaw's research found as much as 41% of some organizations' digital estates is unknown to the central web team, including forgotten microsites, old campaign pages, and stale PDFs with broken links and duplicate titles.
- Offsite content. LLMs tend to look for, and connect to, offsite mentions of your brand, with more than 80% of AI search references (McKinsey) and LLM brand mentions (AirOps) coming from third-party content.
What to Do: Own It, Measure It, Clean It Up
Shaw's prescription is concrete:
- Centralize ownership with one C-suite member — the direct answer to the fragmentation Larridin's data exposes.
- Assess your estate against a maturity model and report to leadership as a simple scorecard, not a technical dump.
- Govern the estate: build an asset register of every digital property and put lightweight approval workflow around new ones.
- Get documents under control. Move from unstructured PDFs to structured, updatable documents.
- Manage offsite content. Be careful what you put out that might get hosted offsite. Send takedown requests for outdated or misleading content hosted offsite.
None of this is glamorous. All of it determines whether AI tells your story or someone else's version of it. (Measuring exactly that is what we do at Larridin, so perhaps we're biased; but the data speaks for itself.)
Shaw's full article adds case detail, an FAQ, and an action table worth stealing.
Read the full article on CMSWire →