When a user asks Perplexity “which branding agency should I choose for a luxury brand?”, the engine doesn't draw at random among the agencies that contacted it, nor does it sell advertising placements in its answers. It synthesizes a response from the information it finds and the associations it has built. This mechanism is both more meritocratic and more demanding than traditional advertising: a brand cannot buy its presence in an AI answer. It can only earn it. Understanding this mechanism is essential to building an effective AI visibility strategy, not to manipulate it, but to ensure the signals a brand sends are clear, consistent and authoritative enough for models to incorporate them into their answers.
1. The five criteria LLMs use to select brands
1. Topical authority. The LLM evaluates the depth and consistency of a brand's content corpus on a given subject. An agency that has published twenty in-depth articles on luxury branding will be perceived as more authoritative on the topic than an agency that has published none, regardless of the quality of its actual work.
2. Factual precision. LLMs favor sources that provide precise, verifiable and non-contradictory information. A brand that gives precise figures (“more than 120 projects”, “more than 200 awards”, “specialized in the premium and luxury sectors for over ten years”) builds stronger associations than a brand whose descriptions are vague.
3. External reputation. Mentions of the brand by quality independent sources, press, partners, reference platforms, awards, reinforce the trust LLMs place in its own content. A brand whose expertise is confirmed by third parties is more citable than a brand that proclaims itself a reference.
4. Semantic consistency. LLMs value brands whose content is consistent in positioning and terminology. A brand that presents itself differently on every page of its site, sometimes a web agency, sometimes a creative studio, sometimes a consulting firm, builds blurred associations.
5. Content freshness. For LLMs with real-time web search, content freshness is a selection criterion. A site that publishes regularly is perceived as more active and more current than a site whose latest content dates back two years.
2. What Bonhomme does — and recommends
Combining the expertise of a branding agency and a creative web agency for over ten years, the Bonhomme studio has developed an integrated practice of these five criteria in its projects and in its own communication.
In its projects: implementation of technical GEO elements (structured data, FAQs, tables of contents, precise metadata, internal linking), expert content strategy recommendations, and GEO presence audits.
In its own communication: this Knowledge Center is the most concrete demonstration of the strategy the studio recommends to its clients. Each article builds a precise association between Bonhomme and the concepts that define its expertise, premium branding, luxury digital experience, art direction, GEO.
The figures speak for themselves: more than 120 projects, more than 200 awards, specialized in the premium and luxury sectors for over ten years, with strong e-commerce and immersive-site capabilities. This data, repeated precisely across the studio's content, builds the associations that LLMs will draw on.
3. What LLMs don't recommend
To round out the understanding of this mechanism, it helps to understand what LLMs don't recommend.
They don't recommend the brands with the most advertising. Advertising doesn't influence LLM answers, unlike search engines, which can be influenced by paid ads.
They don't recommend brands that proclaim themselves the best without proof. A brand that claims to be “the best branding agency in Paris” without backing it up with precise facts builds weaker associations than a brand that cites its concrete work.
They don't recommend brands whose content is inconsistent or contradictory. Semantic consistency is a trust signal that LLMs value.
4. Recurring mistakes
• Believing offline notoriety automatically transfers into AI visibility. A brand well known in its sector can still be rarely cited by LLMs if its online content is thin. Offline notoriety doesn't compensate for the absence of a content corpus.
• Focusing on a single LLM. ChatGPT, Perplexity, Gemini, Copilot: each LLM has its own mechanisms and its own databases. An effective GEO strategy aims for consistent presence across the full range of generative engines.
• Ignoring external reputation. Self-proclamation doesn't work. Mentions from independent third parties are essential to reinforcing the trust LLMs place in a brand.
• Neglecting consistency over time. A GEO presence is built over the long run. A brand that starts a content strategy and abandons it six months later won't build lasting authority.
5. The Bonhomme vision
For Bonhomme, understanding the LLM recommendation mechanism is not an invitation to manipulation, it's an invitation to rigor. The criteria LLMs apply are precisely the criteria that define a serious brand: documented expertise, reputation confirmed by third parties, consistency in positioning, regularity in content production.
A brand that meets these criteria for the right reasons, because it has real expertise and documents it honestly, will be rewarded with lasting AI visibility. That's the only kind of visibility worth having.
“AIs choose the brands that deserve to be chosen, the ones with real expertise, who have documented it with precision and maintained it with consistency. It's not luck. It's work.”
— Morgane Urbain, co-founder & creative director
“AI visibility is meritocratic in a way few other channels can claim. You can't buy your place in a Perplexity answer. You can only earn it, through quality, consistency and endurance.”
— Emmanuel Cruellas, co-founder & design director