BlueSky PR reports two-thirds share of voice in AI searches

BlueSky PR says it captures two-thirds of the share of voice against competing PR agencies in the AI prompts it monitors, as UK small businesses reconsider how customers find suppliers through conversational search.

Managing director Stephanie King describes changes to the agency’s website and content in a guide published by SmallBusiness.co.uk on 14 October 2026. Other contributors warn that appearing more frequently in AI answers does not, by itself, demonstrate stronger marketing performance.

What happened

BlueSky PR has introduced a dedicated frequently asked questions page and an llms.txt file intended to help AI tools interpret its website. King says both are also written to be understandable and useful to people visiting the site.

The agency has added schema and structured data to its principal pages, identifying their subject, author and the company responsible. That work now extends to every new blog post and each older post it updates.

Its service pages begin by explaining the agency’s identity, offer and intended customer. Before making those changes, King says the business checked that conventional search engine optimisation was in place and that nothing prevented AI crawlers from accessing the website.

King also reports that the agency’s own website is the source most frequently cited in the answers it tracks. She contrasts that with research by Muck Rack which, she says, found that earned media accounted for 84% of AI citations rather than brands’ own websites.

The reported share of voice relates to prompts the agency believes its target audience is likely to use. It is not presented as a measurement of every AI search for PR services, and the guide provides no associated sales or enquiry figures for BlueSky PR.

The commercial distinction is between being mentioned and winning business. Separate reporting on Shopify’s comparison of AI referrals and organic search conversions also puts attention on what happens after a potential customer encounters a company through AI.

The background

AI-powered search produces a synthesised response to a question, drawing on information from different sources. Large language models interpret context and meaning, while users often ask longer, more conversational questions than they would enter into a conventional search engine.

Two overlapping disciplines have emerged. Generative Engine Optimisation, or GEO, concerns material that an AI system can summarise and cite. Answer Engine Optimisation, or AEO, focuses on content suited to snippets and other answer-based search features.

Assistants process material in smaller sections, assessing its relevance before assembling a response. A page can therefore contain useful information without supplying a passage that an assistant selects for a particular question.

Another principle described in the guide is co-occurrence: how frequently a business or product appears alongside relevant subjects elsewhere online. Partnerships, guest contributions and participation in relevant forums can help establish those associations beyond a company’s own website.

Conventional SEO remains part of the work. Descriptive metadata, internal links, mobile-friendly layouts and shorter loading times help visitors and search systems navigate a site. The guide advises businesses to address these foundations alongside AI-specific changes, rather than treat GEO as a replacement.

Images and videos provide further context when accompanied by descriptive alternative text or transcripts. Structured data can identify products, reviews and articles, while local business markup supplies machine-readable information about a company.

First-hand accounts, customer testimonials and case studies can distinguish a page from generic material. That emphasis on original experience also connects with the value of freelance judgement in an AI image test: the contribution of a person remains relevant when businesses use AI tools.

What people are saying

James Crawford, managing director of PR Agency One, says small companies should be wary of specialist jargon and promises of rapid results. He describes customers asking assistants to recommend businesses or compare suppliers, rather than carrying out every stage of that comparison across separate websites.

He advises against beginning with content written specifically to appeal to ChatGPT. His preferred approach is to make company information accurate, useful, current and credible wherever customers and AI systems might encounter it. He also warns against list-based spam and prompt-injection tactics.

Crawford’s concern extends to measurement. An answer can vary with the wording of a prompt, the platform, the user, their location and the time of the search. A visibility score based on selected questions cannot reveal the full range of questions other people are asking.

He argues that businesses should examine whether they appear for relevant services, whether their descriptions are accurate and whether independent coverage supports those descriptions. The commercial test is whether that presence contributes to awareness, consideration, enquiries or sales. The focus on reliable information also features in E.ON Next’s approach to customer trust and AI.

James Demetriades, founder of Zaros and Commerce Club, says press features, interviews and expert commentary form a deliberate part of his brands’ visibility strategy. He regards independent coverage as a body of references that helps both prospective customers and AI systems understand the businesses.

On Zaros’s website, he says that means spelling out product materials, dimensions, attributes and positioning. For Commerce Club, the emphasis is on programme structure, eligibility and the results members achieve, rather than leaving visitors to infer those details from promotional language.

King similarly argues that relevant sector coverage helps establish authority. Her account places website improvements alongside third-party references, rather than suggesting that technical changes alone determine whether an assistant recommends a business.

What happens next

For firms reviewing their websites, the guide recommends placing the answer to a likely customer question before supporting explanation. Clear heading levels and separate sections give each subject a defined place, while important information should not be concealed in tabs or expandable menus.

It also recommends short summaries of three or four points where useful, followed by FAQs written in the language customers actually use. Product descriptions should identify specific features and intended users, avoiding broad claims that reveal little about the offer.

Monitoring can begin with questions a business expects customers to ask, or with visibility tools that track mentions and competitors. Some tools also suggest prompts and provide reports on AI-related search features. Those checks offer observations, not a comprehensive measure of the market.

Website analytics can then help identify traffic sources at site or individual-page level, while SEO tools can show whether AI overviews appear for particular searches. Read alongside enquiries and sales, those measures give directors a more practical basis for assessing the work than a standalone count of AI mentions.