Our managing director James Crawford recently contributed to Retail Week’s article on how large language models are changing beauty and wellness shopping. Its central lesson reflects the AMEC GEO Principles: retailers need more than AI visibility. They need credible content and third party endorsement that gives AI and shoppers sound reasons to trust and recommend them.
AI tools are taking on tasks once spread across search results, review sites, editorial content and in-store advisers. Shoppers can use a large language model, or LLM, to interpret ingredients, compare products, assemble a skincare routine or narrow a choice.
That is the backdrop to Retail Week’s feature, How LLMs are changing how consumers shop beauty and wellness, to which our Managing Director and AMEC board director James Crawford contributed. The article is available to Retail Week subscribers.
The piece captures much of the thinking behind the AMEC GEO Principles, which James helped develop. Generative engine optimisation, or GEO, should not begin with a dashboard score. It should begin with what shoppers need to know, the decisions retailers want to influence and the quality of the information available to support an answer.
What AMEC adds is a measurement discipline: document what is tested, repeat the work over time, acknowledge its limitations and connect what appears in AI answers with what people subsequently understand, trust and do.
Start with real shopper questions
A test such as “What are the best beauty brands?” reveals relatively little. A useful prompt reflects an actual need:
- Which moisturiser is suitable for sensitive skin and costs less than £20?
- What evidence supports this product’s ingredient claims?
- Which products would suit a simpler skincare routine for mature skin?
Questions like these reveal the circumstances in which a product might be considered or recommended. Retailers can then examine which brands and products appear, how accurately they are described, which sources support the answer and where important evidence is missing.
The test conditions are important. The model, prompt, date, market and language should be recorded, with outputs saved and tests repeated. One answer from one platform is not a reliable measure of the whole market.
Build evidence
Trust is particularly important in beauty and wellness because claims about ingredients, efficacy and safety can affect people’s health as well as their purchasing decisions.
Clear product pages and ingredient information are essential, but owned content cannot carry the entire argument. AI-generated answers may also draw on independent editorial coverage, qualified expert commentary, clinical evidence, reviews, forums, creator content and regulatory information.
Retail Week rightly highlights the role of earned editorial authority. Creator content can also improve discoverability, particularly where public posts are accessible to search-connected AI tools. However, as James told the publication, “indexable doesn’t mean trusted”.
There is no universal hierarchy that applies to every platform and question. The useful principle is that accurate, current and properly supported information carries more weight than promotional volume.
The objective definitely is the opposite to simply stuffing pages with keywords or flooding the internet with manufactured mentions. It is to create a coherent public record that answers genuine questions and supports claims with appropriate evidence.
Measure the whole journey
AMEC recommends examining three connected areas:
- Upstream reputation – what editorial coverage, experts, consumers, creators, reviews and other public sources say about the brand.
- Search and content readiness – whether reliable information is current, accessible, structured and easy for search engines and AI systems to interpret.
- Downstream AI outputs – whether the brand appears, how it is described, which sources are cited, what is omitted and whether any reputational risks emerge.
The next step is to distinguish grounding, citation, mention, recommendation and referral. None guarantees the next. A source might inform an answer without being displayed. A brand can be mentioned without being recommended. A recommendation might never produce a visit or purchase. This is modern life now and marketers need to deal with it.
AI visibility is therefore useful diagnostic evidence, but it should not be conflated with an outcome. Retailers should connect it with brand and trust research, search behaviour, engaged website visits, product consideration, enquiries and sales wherever the evidence allows.
Retail teams can begin by:
- Building a governed library of real shopper questions.
- Auditing the brand’s public information and supporting evidence.
- Fixing gaps in product content, search readiness and independent validation.
- Testing AI answers transparently and repeatedly.
- Relating those findings to audience and commercial evidence.
PR Agency One combines retail and FMCG expertise, media relations, search and evaluation through its GEO and AI visibility work. We can assess how a brand is represented against real shopper questions, identify the evidence influencing those answers and determine which communication actions are most likely to improve trusted visibility.
