The machine has no memory of your media plan. It remembers your reputation

For decades, marketing operated on a quiet but seductive premise: visibility could be manufactured. Buy enough media, stack enough backlinks, hire the right agency, and the machine would return the favour. Digital search, for all its complexity, was ultimately a system that could be gamed. The arrival of large language models (LLMs) as primary discovery engines is now confronting that premise with something it rarely encountered before: an honest reckoning.

AI search does not rank pages. It synthesises reputation. And reputation, it turns out, is not a campaign deliverable.

This is the uncomfortable truth sitting beneath the growing frenzy around Generative Engine Optimisation, the practice of optimising brand content, citations and structured data to improve visibility within AI-generated answers. As GEO spending accelerates and a new class of consultants emerges to service it, the data quietly tells a more sobering story. According to WARC's 2025 report, The New Rules of Brand Discovery, long-term brand equity accounts for 63% of a brand's visibility in large language model outputs. This significantly outweighs marketing spend (22%), citation volume (11%), and influencer or PR reach (4%). In other words, the majority of what determines whether an LLM recommends your brand today was decided years ago, through every product decision, every customer experience and every earned mention that accumulated before the age of AI search had even begun.

The evidence points to a clear reality: the brands winning AI visibility are not those gaming the system. They are the ones embedding themselves into the moments, conversations and decisions where their category genuinely matters. So intense is the focus on AI discoverability that even categories like FMCG, where LLM-driven discovery has yet to meaningfully scale, are already preparing for the shift.

As Gunjan Khetan, Chief Marketing Officer at Perfetti Van Melle India, points out, FMCG's AI discovery journey is still nascent, given the category's mass nature and lower dependence on high-intent search moments. However, brands are already exploring ways to integrate themselves into AI-led recommendation ecosystems, particularly around recipes and culinary occasions.

The Citation Trap

The GEO industry, in its current form, has largely converged on citation building as its primary deliverable. Agencies measure citation share. Dashboards track mention frequency across ChatGPT, Perplexity, Gemini and Google AI Overviews. The assumption embedded in this approach is that citations drive visibility, that if a brand can engineer enough references in the right places, the LLM will follow.

The data suggests the relationship runs in the opposite direction. A separate study by Ahrefs, analysing 75,000 brands, found that branded web mentions correlate with AI visibility at a coefficient of 0.664, compared to just 0.218 for traditional backlinks, a three to one advantage. Crucially, 82% of all AI citations originate from earned media, and 94% from non-paid sources. You cannot buy your way into citations. The citations are the machine reflecting back what it already believes about your brand, based on the organic conversation that preceded your GEO investment.

Prashant Puri, CEO and Co-founder of AdLift, is direct about the implication. "The accumulated brand equity is already baked into the training data. When the GEO industry treats citation share as its primary metric, it's measuring an outcome, not a cause. The citations are the AI reflecting back what it already believes about your brand. The real work is building the underlying authority that produces those citations, press coverage, third party mentions, expert content published in credible places. GEO tactics can help at the margins, but they can't shortcut that foundation."

This matters enormously for how marketing budgets are being allocated. If 63% of LLM visibility is already determined by historical brand equity, the industry spending against the remaining 11%, citation volume, while neglecting long term brand building is, arithmetically, inverting its own priorities. Rahul Khanna, Founder of BarCode, frames it plainly. "My worry is we're about to watch marketers spend serious money optimising the 11% while starving the 63%."

Khanna draws a pointed parallel to a previous era's mistake. "Chasing citation volume without the underlying brand is the 2026 version of buying followers." What makes this cycle particularly risky is that the investments feel productive in the short term. Citation dashboards move. Metrics tick upward. The bill gets paid. The brand, meanwhile, is not getting built.

Strong Brands, Not Big Brands

If brand equity is the dominant variable, the natural anxiety is that AI search simply entrenches incumbency, that the giants with decades of earned presence will compound their advantage while challengers find the door permanently closing. The reality, as the data shows, is more nuanced and, for the right kind of emerging brand, more encouraging.

The critical distinction is between large brands and strong brands. These are not the same thing, and the difference is precisely what AI search exposes.

Ajay Verma, Co-Founder and Managing Partner of 0101.Today, puts it squarely. "AI doesn't automatically favour size. It favours trust and relevance. Brands like Zoho and Zerodha are frequently surfaced because they've built authority through education and customer value over time. GEO is ultimately a brand building exercise, not a citation building exercise." He adds that the winners in AI search will be the most useful brands, not necessarily the biggest ones.

The Ahrefs December 2025 follow up study introduced a new variable that reinforces this argument. YouTube mention impressions correlated with AI brand visibility at approximately 0.737, outperforming every other measured signal. A brand that has built a genuine, engaged audience on YouTube is, by that metric, more likely to surface in AI answers than a brand with a superior backlink profile and a larger media budget. The model rewards authentic, repeated, third party engagement, not scale alone.

For challenger brands, the strategic implication is specific. Puri describes it as a shift from breadth to density. "The playbook cannot be about chasing incumbents on raw scale. The strategy must shift from broad keyword volume to category density, being the most consistently cited brand within a hyper specific niche conversation. By dominating third party mentions, expert forums, and specialised vertical media, you build high entity authority within a tight cluster. The AI recognises that contextual relevance."

Khanna takes the argument further, locating the real vulnerability not in small brands but in large, undifferentiated ones. "AI search doesn't favour big brands, it favours strong ones, and those are not the same thing. The brands that should be losing sleep aren't the small ones. It's the large, undifferentiated ones that coasted on media spend for a decade. WARC says that spending now buys you 22% of the outcome. The other 63%, you have to deserve."

The FMCG Variable and the Earned Credibility Imperative

As mentioned earlier, not all categories face AI discoverability equally, and the conversation would be incomplete without acknowledging where the structural limits lie. Gunjan Khetan offers perspective grounded in the realities of mass consumer goods. "Unfortunately, FMCG has not yet scaled in AI discovery, because of the mass reach and the moments not being critical enough to require a massive engagement. The impact of LLMs has been significantly less." But he identifies a meaningful entry point. "Some of the work we're doing today is to integrate our brands in recipes and culinary moments and try to get recommended as part of a larger recipe system."

This points to a broader principle. The brands winning AI visibility are not the ones gaming the system. They are the ones embedding themselves into the moments, conversations and decisions where their category genuinely matters. For travel, hospitality and considered purchases, those moments are abundant and the AI opportunity is immediate. For FMCG, the pathway runs through culinary content, usage occasions and utility driven editorial, not citation engineering.

Khetan is clear about the sequencing. "The first thing will have to be creating credible content that LLMs can pick up. GEO comes much later. If you do a good job at creating credible content that LLMs can pick up, GEO is less of a worry." The implication aligns with what the WARC data shows. Credibility precedes discoverability, not the reverse.

Tools that help marketers understand and measure the gap between what an LLM says about their brand and what they want it to say are becoming increasingly relevant. AdLift's AI Search Visibility platform Tesseract, for instance, recently launched a Claude AI integration alongside AI Traffic Analytics, enabling marketing teams to analyse the context, sentiment and intent behind AI generated brand mentions across platforms, and to connect AI citations to actual traffic outcomes. As the discipline matures, the ability to diagnose where brand authority is underrepresented in AI systems, and to build toward it methodically, will separate credible strategy from dashboard theatre.

What AI Search Is Actually Testing

Beneath the tactical questions of GEO, a more fundamental shift is underway in how brands are discovered, evaluated and chosen. AI search is not simply a new channel. It is, as Khanna describes it, "the great audit of Indian marketing. Every brand that bought visibility instead of building it is about to find out the difference."

The structural change is this. For most of the digital era, the gap between a brand's actual standing and its perceived standing could be managed with money. Enough media spend, enough SEO investment, enough influencer activation could keep the gap hidden. LLMs, trained on the accumulated organic conversation of the web, have less patience for that gap. They synthesise what independent sources actually say, consistently, over time. Verma captures it cleanly. "AI is not changing the fundamentals of marketing. It is making them more important."

The 37% of LLM visibility that is not predetermined by historical brand equity, the window that remains open to real time influence through content freshness, earned media and targeted editorial, is where the near term battleground sits. For challengers with a clear point of view and the discipline to execute consistently, that window represents a compounding advantage. For incumbents resting on legacy spend, it represents a quietly closing door.

The GEO industry will continue to evolve, and citation building as a practice is not without merit at the margins. But the data is now unambiguous about where the majority of AI discoverability is won or lost. It is won in brand building decisions made years before the algorithm is consulted. It is lost in the comfortable belief that the machine can be optimised into trusting you.

In the age of AI search, the most important marketing question is not how to be seen. It is whether, over years of consistent, credible, independently verified presence, you have given the machine a reason to believe you.