The brand is no longer in the room

A customer in Munich is deciding whether to buy your product. She has questions about quality, suitability, how it compares to alternatives. She will make a decision within the next few minutes. You are not part of this conversation. You will never know it happened.

She is not on your website. She is not in your store. She is asking an AI assistant, in German, waiting for the kettle to boil. The AI is answering based on what it knows - which is based, in part, on content you published. Perhaps your product descriptions. Perhaps a buying guide on your site. Perhaps nothing of yours at all.

The AI will either recommend your product or it won't. It will either represent your brand accurately or it won't. You have no opportunity to intervene, to clarify, to sell. The conversation is happening without you.

This is new. Not theoretically new, not new in some speculative future sense, but happening now, billions of times daily, across every market and product category. It changes what content must accomplish in ways that most brands have not yet confronted.

The end of the controlled experience

For decades, brand strategy has been built on the principle of controlling touchpoints. The website is designed to create a particular impression. The packaging reinforces it. The store environment, the advertising, the customer service scripts - each element is crafted to ensure the customer experiences the brand as intended.

This worked because the brand was always in the room. When a customer visited the website, the brand controlled what she saw, in what order, with what emphasis. When she walked into a store, the brand controlled the lighting, the music, the way products were displayed. The customer's experience was, to a meaningful degree, the brand's to shape.

AI-mediated discovery breaks this model. The customer asks a question; the AI provides an answer. The brand's content may have informed that answer, but the brand did not deliver it. The AI did - in its own voice, according to its own logic, with whatever emphasis it deemed appropriate.

The brand was not in the room.

What happens in your absence

Consider what occurs when an AI assistant fields a query about your product.

The AI draws on information it has encountered: product descriptions, reviews, editorial content, specifications, comparisons. It synthesises this into a response calibrated to the customer's question. It may quote. It may paraphrase. It may summarise. It may simply assert, without attribution, that your product is or is not suitable for the customer's needs.

Your carefully crafted brand voice - the rhythm, the vocabulary, the personality you have cultivated - is unlikely to survive this process. The AI is not trying to represent your brand. It is trying to answer a question. If your content helps it do so, that content will be used. But it will be used on the AI's terms, not yours.

This is not a failure of the technology. It is simply what happens when an intermediary speaks on your behalf. The intermediary has its own way of speaking.

The content you leave behind

There is an uncomfortable implication here. In AI-mediated discovery, your content functions less like a sales conversation and more like a briefing document. You are preparing someone else - something else - to speak about your products in conversations you will never witness.

This reframes what content must accomplish. The question is not only "will this persuade a customer who reads it?" but also "will this equip an AI to represent us accurately when we are not present?"

These are different questions. Content that persuades through atmosphere, through accumulated effect, through the slow build of brand impression, may not transfer well to an AI intermediary. The AI cannot replicate atmosphere. It extracts information.

Content that performs well in this context tends to be substantive: specific about what makes a product distinctive, clear about who it suits and who it doesn't, honest about trade-offs, rich in the kind of detail that allows an informed recommendation. Research from Princeton University demonstrates that content featuring statistics, citations and quotations can increase visibility in AI responses by 30-40%. This is not incompatible with brand voice, but it requires that brand voice be grounded in substance rather than style alone.

The multilingual dimension

This dynamic is complicated further by language.

Your customer in Munich is speaking German. The AI is responding in German. AI platforms now support dozens of languages - Perplexity alone operates across 46 - and the content informing responses is predominantly drawn from the language of the query.

If your German-language content is thin, or translated in ways that feel mechanical, the AI has less to draw upon when representing your products to German-speaking customers. It may rely instead on competitor content that is richer, more natural, more authoritative. Your products may be mentioned; they may also be overlooked entirely in favour of alternatives the AI understands better.

The brands you compete with in Germany are not only those with better products or stronger marketing. They are those whose German-language content better equips AI systems to understand and recommend their offerings.

This is true in every market, in every language. The AI-mediated conversation is happening in Japanese, in French, in Portuguese, in Dutch. In each case, the quality of your content in that language shapes whether and how your products are represented.

The question of trust

There is something vertiginous about this situation. Brands have invested enormously in controlling their narratives, and now those narratives are being retold by systems they do not control, in conversations they cannot see, to customers they may never meet.

Some will find this threatening. It is, in a sense. The comfortable assumption that a customer's journey passes through brand-controlled territory is no longer reliable. Sixty-six percent of frequent online shoppers now use AI assistants to guide purchase decisions. ChatGPT has become the second-most popular source of product recommendations, trailing only physical stores. The customer may arrive at a purchase decision having never encountered the brand directly.

But there is another way to view this. The AI is not hostile. It is not trying to misrepresent your products. It is trying to be helpful to the person asking. If your content genuinely helps it do that - if your product descriptions are clear, your specifications accurate, your differentiation well-articulated - the AI becomes an informed advocate rather than an unreliable narrator.

The question is whether you have given it enough to work with.

The owned experience still matters

It would be easy to read this situation as a call to abandon the human reader entirely - to optimise exclusively for AI extraction, treating content as machine fodder rather than brand experience. This would be a mistake.

The reality is more nuanced. AI-referred traffic still accounts for less than one percent of total e-commerce sessions for most retailers. The vast majority of customers still arrive through traditional channels: direct navigation, organic search, email, social media. These customers experience your content directly, unmediated by AI interpretation.

Moreover, research indicates that AI-referred visitors currently convert at lower rates than those arriving through established channels. They browse, they research, but they often return later through direct navigation to complete their purchase. The owned experience - your website, your product pages, your checkout flow - remains where conversion actually happens.

Content must therefore serve two masters simultaneously. It must be substantive enough to brief an AI intermediary effectively. But it must also be compelling enough to persuade a human reader who arrives directly, who experiences your brand voice firsthand, who makes a decision based on how your content makes them feel as much as what it tells them.

This is not a contradiction. Substantive content that respects the reader's intelligence, that provides genuine insight rather than empty claims, tends to perform well for both audiences. The AI extracts the facts; the human appreciates the clarity. What fails with both is content that is neither informative enough for AI extraction nor distinctive enough for human engagement - the hollow marketing copy that satisfies no one.

The shift is accelerating

The numbers suggest this is not a passing curiosity. A Pew Research study found that when Google displays AI-generated summaries, users click through to websites roughly half as often - 8% compared to 15% without the summary. The information is consumed; the visit never occurs. The brand's website sits unused while the customer's question is answered elsewhere.

This is not limited to Google. Every AI assistant, every chatbot, every voice interface that answers product questions is drawing on content created by brands - and delivering conclusions without attribution, without context, without the careful framing that brands have spent decades perfecting.

What this demands

The instinct, for many brands, will be to find ways to reassert control. To optimise for AI citation the way they once optimised for search rankings. To treat this as a technical problem requiring a technical solution.

This is not wrong, exactly, but it may miss the more fundamental point. The change is not primarily about tactics. It is about accepting that the brand is no longer always in the room - and ensuring that what you leave behind, in every language and every market, is sufficient for others to represent you well.

This means content that informs AI intermediaries without abandoning the humans who still arrive directly. Content that is rich enough in substance to survive extraction, yet distinctive enough in voice to reward engagement. Content that works whether it is read in full on your website or summarised by an algorithm in a kitchen in Munich.

This is not a comfortable position for organisations accustomed to controlling their narratives. But it is the position nonetheless.

The kettle has boiled. She has made her decision. You were not consulted. The question is whether the content you created, months or years ago, in a language you may not speak, served you well in a conversation you will never know occurred.

Time for tea.