AI is much bigger than you think. Most brands are still underestimating what that means.

For the last two years, a lot of the conversation around AI has sat in the wrong place. Too much of it has been framed as theatre. Hype. Future talk. A bit of experimentation on the side while the “real” work still happens in Google, Amazon, Meta, and retailer search. That framing is now badly out of date. New research from [Graphite](https://graphite.io/five-percent/ai-is-much-bigger-than-you-think) suggests AI usage is already far larger than most marketers and commerce teams think. Their analysis argues that AI now generates roughly 45 billion monthly sessions worldwide and that total AI usage is already 56% the size of traditional search globally. Even when narrowed to search-like “asking” behaviour, AI usage is still 28% the size of search worldwide. Perhaps most importantly, the shift is not just happening on desktop websites. Around 83% of AI usage worldwide is happening inside mobile apps. That matters because many brands are still looking at the market through a web-first lens. They compare Google web traffic with ChatGPT web traffic and conclude that AI is still niche. But that is the wrong comparison. It misses app usage. It misses non-Google search behaviour. It misses the fact that discovery is now spreading across multiple AI interfaces, not one browser tab. Graphite’s core point is that older estimates have understated AI’s true size by around 4x to 5x because they excluded app usage. For brands, the practical conclusion is simple. This is no longer a future channel. It is already a live discovery layer. And if your team still treats AI visibility as a side project, you are probably late. The biggest mistake is thinking AI and search are zero-sum A lot of commentary still assumes one clean narrative: if ChatGPT rises, Google falls. If AI grows, search dies. If consumers ask LLMs for advice, they stop using search engines. The more useful interpretation is messier and more commercially relevant. The pie is getting bigger. Graphite argues that search engine usage has not collapsed and that overall search-related activity across search engines plus AI has increased, not decreased. Their estimate is that combined search usage grew 26% worldwide from Q1 2023 to Q4 2025. That matches what many of us are seeing in the market. Consumers are not replacing one behaviour with another in a neat line. They are layering behaviours. They still Google. They still search on Amazon. They still use TikTok, Reddit, YouTube, and retailer sites. But now they also ask ChatGPT what to buy, ask Perplexity for comparisons, use Gemini for guided discovery, and increasingly move between research surfaces without thinking much about the platform boundary. For brands, that means the strategic question is not “Will AI replace search?” It is “How do we make sure we are visible wherever discovery is expanding?” That is a much better question. It is also a harder one. AI discovery is not just a media problem. It is a content and structure problem. Many companies still instinctively respond to new channels by asking the media question first. Should we advertise there? Can we buy placement? Is there a beta product? What is the CPC? Those questions matter. But they are not the first ones. The first battle in AI discovery is organic. Large language models do not pull brand preference out of thin air. They build answers from what they can interpret across the open web, product detail pages, structured data, reviews, forums, marketplace content, editorial mentions, and brand-owned assets. They then synthesise those sources into recommendations, comparisons, and summaries. That means your visibility in AI systems is heavily shaped by how well your brand is understood. Not just whether you exist. Whether you are legible. Can the model easily understand what your product is, who it is for, what problem it solves, how it differs, what attributes matter, what proof exists, and where it sits in the consideration set? That is why so much traditional brand and eCommerce content underperforms in AI environments. It was written for channels that rewarded a different kind of optimisation. Search wanted ranking signals. Marketplaces wanted conversion signals. Paid social wanted thumb-stopping hooks. AI discovery wants something else as well: clear, structured, machine-readable commercial meaning. Brands that win in this environment tend to do a few things better than everyone else. They make product claims more explicit. They cover use cases in plain language. They answer comparison questions before customers ask them. They create consistency across DTC, Amazon, retailer, and editorial surfaces. They use content architecture that helps models connect attributes, audience, and outcomes. This is not just SEO with a new label. It is broader than that and in many ways more commercial. Why this matters especially for challenger brands The interesting part is that this shift may not favour the biggest brands as much as people assume. In traditional search, incumbency often carries massive advantages. Bigger sites, stronger link profiles, more branded demand, deeper media budgets. In AI search and agentic discovery, some of those advantages still matter. But they are not the whole game. If a challenger brand is clearer, more consistent, better reviewed, more sharply positioned, and more semantically rich across its content ecosystem, it can outperform larger competitors in the kinds of prompts that matter. That is particularly true in high-consideration and attribute-led categories. Think haircare devices. Sports nutrition. Supplements. Sunscreen. Footwear. Oral care. Functional wellness. Any category where consumers ask layered questions like: What’s best for sensitive skin? Which running gel is easiest on the stomach? What’s the best hairdryer for reducing damage? What is similar to X but cheaper? Which mouthguard is best for braces? What’s a good daily multivitamin for men over 40? These are not just keywords. They are commercial conversations. And AI is increasingly where those conversations start. **The mobile point is bigger than most marketers realise** One of the most important takeaways from the Graphite research is also one of the easiest to miss. AI usage is heavily app-led. Globally, 83% of AI sessions in their analysis happen in mobile apps. In the US, it is 75%. That should force a reset in how brands think about traffic and discovery. Many reporting stacks still over-index on web analytics and website referrals. But if discovery starts in an app-based AI environment and ends in a retailer, marketplace, or offline purchase journey, the old attribution pathways get weaker. That does not mean the channel is unimportant. It means measurement is lagging behaviour. The teams that wait for perfect attribution before acting will move too slowly. This is the same mistake many businesses made in earlier platform shifts. They waited until the dashboard was clean and the causal chain was obvious. By then, the advantage had already been taken. **So what should brands do now?** First, stop treating AI visibility as abstract innovation work. This is now a commercial capability. It sits alongside SEO, Amazon optimisation, retailer content, and paid media. It should be part of the growth conversation, not parked in a trends deck. Second, audit how your brand currently appears in AI-driven discovery. Not just whether you are mentioned. How you are framed. Which competitors appear instead. Which attributes get associated with your products. Which use cases you win. Which ones you miss. Third, strengthen the source material. That means better PDPs. Better category language. Better structured content. Better FAQ coverage. Better comparison framing. Better consistency between brand site, Amazon, retailers, and third-party surfaces. Fourth, focus on commercial prompts, not vanity prompts. A lot of brands waste time checking whether ChatGPT knows their brand name. That is not the real test. The real test is whether you appear when a customer describes the problem, need state, category, or desired outcome without naming you. That is where growth sits. Fifth, connect organic AI visibility to retail execution. If a model recommends you but your Amazon PDP is weak, your title is vague, your bullet points are generic, your images do not reinforce the use case, or your reviews do not support the claim, you waste the opportunity. AI visibility without conversion infrastructure is incomplete. Conversion infrastructure without AI visibility is becoming less sufficient. The two now need to work together. What this means for Lmo7 At Lmo7, we think this shift is pointing toward a bigger change in commerce. Search is no longer one place. Discovery is no longer one format. Optimisation is no longer one discipline. Brands need to become discoverable across an expanding network of AI, search, marketplace, and agentic surfaces. That requires a tighter operating model between content, commerce, and intelligence. The winners will not just be the brands with the biggest budgets. They will be the brands that become easiest for machines to understand and easiest for consumers to choose. That is the real game now. AI is much bigger than most people think. And the commercial implications are bigger still.