AI Citations Aren’t Recommendations: My Interview with Majestic on Brand Visibility in 2026
SEO in 2026: Additional Insights | Maryanna Franco
Being cited by AI doesn’t mean your brand is being recommended.
A brand can appear frequently in AI-generated answers, have substantial topical authority and publish extensive content, yet still be absent when someone asks ChatGPT or another AI assistant which product, service or company to choose.
That distinction was at the centre of my interview with Majestic for its SEO in 2026: Additional Insights series.
We discussed why citation counts alone can create a misleading picture of AI visibility, how brand associations influence recommendations, and why publishing more content isn’t always the answer.
Watch, listen to or read my interview on Majestic
Why AI citations don’t tell the whole story
For years, SEO professionals have measured visibility through rankings, impressions, clicks and organic traffic.
With the emergence of AI search, we’ve added new metrics: AI citations, brand mentions and share of voice.
These are useful measurements, but they don’t all represent the same outcome.
Consider three different situations:
- Citation: An AI assistant uses your website as a source when answering a question.
- Mention: Your brand appears by name in an AI-generated answer, with or without a citation.
- Recommendation: Your brand is presented as a suitable option when someone is evaluating what to buy, use or hire.
A company might perform well in the first two categories and struggle in the third.
This matters because commercial visibility isn’t simply about being present in an answer. It’s about being considered when someone is making a decision.
As I explained during the interview, being part of AI’s bibliography isn’t the same as being on its shortlist.
Citations remain valuable. They can tell us whether our content is being discovered and used as supporting material. But they aren’t a reliable substitute for measuring recommendations.
The missing link: brand and category associations
One of the most important questions in AI visibility is not simply whether an AI system recognises your brand.
It’s whether that system associates your brand with the products, services, attributes, audiences and use cases you want to be known for.
Imagine a company selling running shoes.
An AI assistant might recognise the company as a sportswear brand. It might cite its website in answers about running equipment.
But what happens when someone asks:
Which running shoes are best for wet weather?
The assistant now needs to connect the brand to a much more specific context: running shoes, wet-weather performance, grip, durability and suitability for particular conditions.
Being recognised as a sportswear company doesn’t automatically establish all those relationships.
This is where brand association becomes important.
Why co-occurrence matters
Co-occurrence describes how frequently and in what contexts entities or concepts appear together.
For brand visibility, I’m particularly interested in the relationships between:
- A brand and its primary category.
- A brand and its products or services.
- A brand and the audiences it serves.
- A brand and the problems it solves.
- A brand and the attributes that distinguish its offering.
These relationships need to be communicated clearly on the brand’s own website, but also through relevant external sources.
For example, an article that mentions a running-shoe brand in the context of wet-weather footwear provides more specific category information than an unrelated mention of the company’s name.
That doesn’t mean every co-occurrence will directly influence an AI recommendation. AI systems have different retrieval and generation processes, and recommendations depend on more than entity associations.
But examining those relationships can help explain why a brand is recognised in one context and overlooked in another.
Why publishing more content isn’t always the solution
Another topic we explored was what I call the content volume trap.
When a brand isn’t appearing in AI answers, the instinctive SEO response is often to publish more content.
More articles. More landing pages. More FAQs. More keyword coverage.
Sometimes that’s exactly what’s needed.
But not always.
If an AI system already recognises your brand, understands what you sell and regularly uses your content as a source, producing another 50 articles might not solve your recommendation problem.
The issue could instead be how the brand is positioned relative to its competitors.
Are other companies consistently described as leaders or specialists in a particular category?
Are they mentioned in relevant industry publications, comparison articles, interviews or podcasts?
Do independent sources connect them with specific use cases more clearly than they connect your brand?
These are different problems from insufficient content coverage, and they require different interventions.
Before recommending more content, I want to understand exactly where visibility is breaking down.
How I measure AI brand visibility
In the Majestic interview, I described an approach based on testing three distinct dimensions: recognition, citations and recommendations.
Rather than relying exclusively on an aggregated AI visibility score, I want to understand how a brand performs across different types of queries.
1. Test brand recognition
Start with brand-dependent prompts:
- What is [Brand]?
- What does [Brand] do?
- What products or services does [Brand] offer?
- Which category does [Brand] belong to?
The objective is to understand whether AI systems identify the brand and its commercial positioning accurately.
2. Test informational visibility
Next, investigate whether the brand appears as a source or is mentioned in answers related to its expertise, products and use cases.
For a running-shoe company, this could include questions about footwear materials, running conditions, durability or traction.
Here, I examine whether the brand is cited, whether it is mentioned, and the context in which it appears.
3. Test commercial recommendations
Finally, remove the brand name from the prompt.
Ask questions such as:
- What are the best running shoes for wet weather?
- Which brands specialise in running shoes with good grip?
- What running shoes would you recommend for someone running in rainy conditions?
This is where we investigate whether the brand makes the shortlist without being explicitly requested.
The distinction between brand-dependent and brand-independent testing is particularly important.
If I ask an AI assistant about my company by name, I may learn whether it recognises me. I haven’t necessarily learned whether it would recommend me to someone who doesn’t already know I exist.
For meaningful comparisons, these tests need consistent prompts, clearly defined categories, repeated observations and careful documentation of the conditions under which responses were generated.
What I investigate when a brand is recognised but not recommended
This is where the diagnostic work becomes particularly interesting.
If a brand has strong recognition and informational visibility but weak recommendation visibility, I start investigating the evidence around the brands that are being recommended.
Where do they appear?
How are they described?
Which categories and attributes are associated with them?
What do independent publications say about them?
Are those relationships also present in descriptions of our own brand?
I then compare three layers of evidence:
What the brand says about itself. Its website, product pages, service pages, positioning and content.
What the wider web says about the brand. Industry publications, relevant third-party mentions, interviews, comparisons and other independently controlled sources.
What AI systems say about the brand. How they categorise it, which attributes they associate with it, when they cite it and whether they recommend it.
The differences between these three layers can reveal useful diagnostic opportunities.
For example, a company might position itself as a specialist in a particular service, while independent sources consistently describe it using a broader category.
Or competitors might be repeatedly associated with a valuable use case that is barely mentioned in connection with the company being investigated.
Neither observation proves a specific cause of an AI recommendation gap. But both provide evidence worth investigating before deciding what to change.
What this means for SEO and GEO strategy
I don’t see Generative Engine Optimisation as a reason to abandon SEO.
Technical accessibility, useful content, clear website architecture and search visibility remain important.
But the scope of our work is expanding.
We need to consider not only whether search engines and AI systems can retrieve information about a brand, but also whether the information available across different sources communicates a coherent and accurate picture of what that brand does.
That means SEO, content strategy, digital PR and brand positioning are increasingly connected.
It also means our recommendations should begin with diagnosis rather than a predetermined checklist.
Sometimes the right intervention is improving a product page.
Sometimes it’s clarifying how a service is described.
Sometimes it’s strengthening relevant third-party relationships or correcting inconsistent brand information.
And sometimes the evidence suggests that the brand is already well understood, but isn’t sufficiently differentiated in the commercial context being tested.
The solution depends on the problem.
My main takeaway from the Majestic interview
If I could leave marketers with one recommendation, it would be this:
Don’t stop measuring AI visibility. Start measuring what that visibility actually means.
A citation tells you something about source usage.
A mention tells you something about brand presence.
A recommendation tells you something about whether your brand is being considered as a potential choice.
Those are different observations, and treating them as interchangeable can lead to the wrong strategic decisions.
The question isn’t simply how often AI talks about your brand.
It’s whether AI understands what your brand does, which category it belongs to, who it’s relevant for, and when it should be considered.
That’s where I believe some of the most valuable diagnostic work in AI visibility now lies.
Watch or listen to the original Majestic interview
I discussed these ideas in more detail as part of Majestic’s SEO in 2026: Additional Insights series, including brand-category relationships, co-occurrence, AI visibility measurement, query fan-out and how to prioritise interventions.
Interview: Test whether AI associates your brand with your category, not just whether it cites you
Publication: Majestic — SEO in 2026: Additional Insights
Featured expert: Maryanna Franco, SEO & AI Visibility Consultant at BrilliantSEO
Watch the interview, listen to the podcast or read the full Q&A on Majestic
Bring this approach to your marketing team or event
I work with marketing teams and organisations to understand how search engines and AI systems discover, interpret and recommend brands.
Through corporate workshops, training and conference presentations, I explore the relationship between SEO, GEO, brand associations and commercial visibility, using practical diagnostic methods rather than generic optimisation checklists.
Interested in a workshop or conference session?
Explore my corporate training and speaking opportunities, or get in touch to discuss your event or team’s needs.
About Maryanna Franco
Maryanna Franco is an SEO and AI Visibility Consultant and founder of BrilliantSEO. Her work focuses on technical SEO, international SEO, content strategy, conversion-oriented search strategies and how brands are recognised and recommended in AI-powered search experiences.