One billion people now use Google AI Mode every month. They're not clicking on blue links — they're asking questions and receiving synthesized answers that name some brands and skip others, often completing a purchase before the recommended brand's website ever enters the picture. Most brands have no reliable way to know whether they're in those answers or out of them.

That gap between what AI platforms say about a brand and what that brand can actually measure is not a niche marketing problem. AI Overviews in standard Google Search already correlate with a 61% drop in organic click-through rates and a 68% drop in paid click-through rates. Fortune estimates AI agents account for roughly 10% of revenue at some leading brands. And 57% of consumers say they'd let an AI agent switch them to a different brand for a better-value product — which means the brand losing that customer often doesn't know who took their place, or that a sale was ever possible.

On July 2, a company called Profound launched Aim: an autonomous background agent for marketing teams that continuously monitors AI search signals across ChatGPT, Perplexity, Gemini, and Microsoft Copilot, then converts what it finds into prioritized marketing actions and execution workflows. Profound had already raised $96 million in a Series C led by Lightspeed Venture Partners — with Sequoia and Kleiner Perkins participating — reaching a $1 billion valuation roughly 18 months after founding. About 10% of the Fortune 500 are now clients.

That last figure carries a specific signal. Three of the most discerning venture firms in technology put $96 million into the problem of knowing whether ChatGPT recommends your brand. Capital at that scale from those particular firms means the AI citation gap has crossed from experimental marketing concern to something CFOs and boards are starting to treat as infrastructure risk. The brands that understand why — and what to do about it — will be in a fundamentally different position from those that don't when the 2026 holiday season arrives.

What Profound Aim Does — and Why It's Different from Dashboards

Most brand-side responses to the AI citation problem have been passive: set up monitoring tools, check a report weekly, notice that Perplexity cited a competitor, file a note to update some content. The data comes in; human teams decide what to do with it; weeks pass.

Profound Aim works differently. It monitors visibility, sentiment, and accuracy across AI platforms — ChatGPT, Perplexity, Gemini, Microsoft Copilot — and identifies which signals represent opportunities significant enough to act on. It then converts those opportunities into structured marketing "Projects" with briefs, tasks, and execution workflows, and routes work to specialized Profound Agents for research, content creation, and optimization. Human teams approve and oversee; Profound tracks how completed work shifts the underlying AI visibility metrics.

The closed-loop structure is the key architectural difference. A traditional analytics platform tells you what happened. Aim tells you what to do next and then measures whether it worked.

Plaid's Organic Growth Manager, evaluating the early product, described the surfaced projects as "literally gold." Profound's current clients include Figma, Walmart, Ramp, MongoDB, Chime, and U.S. Bank. Five hundred or more customers are running Profound Agents daily, across a base of over 700 enterprise clients.

Lightspeed's Sachin Patel, in remarks accompanying the Series C, framed the company's positioning in explicitly infrastructure terms: Profound Agents position the company to "define how marketing is done in an agentic world." The framing echoes how Stripe was described circa 2013 — not as a product but as rails. The bet is that measuring and acting on AI citation data will eventually be as fundamental to enterprise marketing operations as tracking conversion rates is today.

Why Brands Are Disappearing

The AI citation gap isn't random. There are structural reasons why certain brands disappear from AI-generated answers, and those reasons illuminate exactly what the new competitive environment rewards.

AI recommendation engines don't operate on the same signals as traditional search. They don't follow backlinks, keyword density, or paid placement. They are trained on large bodies of text and respond to queries by synthesizing information from sources that have established authority, clarity, and trust in their domain. A brand that has maintained strong coverage in trade publications, published original data that gets cited, maintained clean and structured product information, and accumulated positive reviews across multiple platforms will surface more reliably than one that has optimized for Google PageRank but never published an original study.

Structured data is unusually important. AI agents need to ingest product information in machine-readable formats — live inventory, real-time pricing, delivery commitments, nutritional data, sizing guides — without navigating the friction of a human-oriented webpage. Merchants who have invested in this layer for reasons unrelated to AI (for Google Shopping feeds, for third-party aggregators) find themselves ahead. Those who haven't are effectively invisible to agents that need to make specific recommendations fast.

The measurement problem compounds the visibility problem. If a brand can't see how often it's appearing in AI answers, it can't identify where it's losing ground, can't test whether content changes are shifting AI behavior, and can't detect when a competitor's product gets recommended instead of theirs in ten thousand simultaneous conversations.

The Brand Loyalty Data

The stakes are not hypothetical. Several major research efforts published in the past few months have converged on the same finding: AI agents are not brand loyal, and consumers are largely comfortable letting them switch.

Checkout.com's report, published June 9 and based on surveys across six markets, found that 57% of consumers would allow an AI shopping agent to switch brands for a better-value option. The same survey found that 89% of merchants are actively preparing for agentic commerce, and 72% agree that consumers will adopt agent-led shopping faster than most merchants are ready for.

Deloitte's 2026 Retail Industry Global Outlook puts the executive-level alarm even higher: 81% of retail executives believe generative AI will erode brand loyalty by 2027. Roughly half anticipate the collapse of the traditional multi-step customer journey — replaced by a single AI-mediated interaction — within the same period.

Bain's March 2026 analysis, produced in collaboration with Stripe, found that 30–45% of US consumers already use generative AI to research and compare products. The Bain report is unusually pointed about the merchant risk: retailers that cede checkout control to third-party agent platforms lose the customer data that informs retargeting, replenishment, and repeat-purchase strategies. The result, in Bain's framing, is a retailer that becomes "little more than a fulfillment pipe" — operationally capable of shipping goods but stripped of the direct customer relationship that builds pricing power and loyalty over time.

Kartik Hosanagar, a Wharton professor whose research covers algorithmic commerce, put the risk more directly in January: retailers ceding customer interactions to AI platforms risk becoming mere "fulfillment" operations, stripped of brand experience entirely.

The data suggests the consumer side of this transition has already happened faster than most merchants expected. Agents optimize for objective criteria — price, delivery speed, ratings, availability — over brand familiarity. A consumer who would never have considered switching from their usual brand of running shoes after fifteen years of loyalty may never think about it at all if their AI agent simply returns the best-value option and completes the purchase.

The New Discipline: GEO and AEO

The marketing response to this structural shift has been the rapid development of two new disciplines, which the industry has settled on calling Generative Engine Optimization (GEO) and Answer Engine Optimization (AEO).

GEO focuses on building brand authority and credibility so AI systems cite and recommend a brand within generated responses. The practices are related to but distinct from traditional SEO: rather than chasing keyword rankings, GEO focuses on entity authority, original data publication, structured data cleanliness, expert coverage in trusted sources, and community presence in forums and communities that large language models treat as reliable signals. The goal is "share of answer" rather than "position in search results."

AEO focuses more narrowly on ensuring that AI agents can answer specific factual questions about a brand accurately and completely: product specifications, delivery commitments, return policies, pricing, compatibility. If an AI agent asks whether a merchant ships internationally before making a purchase recommendation, and the answer isn't cleanly available in machine-readable form, the merchant loses that recommendation to a competitor whose data is cleaner.

Microsoft published a detailed playbook on both disciplines in January 2026, titled "From Discovery to Influence," providing practical guidance for structuring product data, schema, and content for AI-driven discovery, agents, and agent-led purchases. Google's guidance for Merchant Center similarly now includes instructions for AI Mode visibility.

The caveat, as Fortune's January 2026 coverage noted, is that GEO "is still more art than science." AI models handle product queries relatively well but are "inconsistent and error-prone when asked questions that pertain to a company's financial stability, governance, and technical certifications." Models can double down on errors when prompted to verify them. Results can shift "from one minute to the next." Companies that invest heavily in GEO based on the assumption that AI responses are consistently deterministic may find the ground moving underfoot.

AIVO Standard, a company founded by Tim de Rosen, has built both a GEO methodology and a system for tracking and governing what information AI agents actually use in their decision-making — addressing the gap between what a brand believes AI is saying about it and what AI is actually saying. This governance layer, still nascent, may become as important as the optimization layer if regulatory scrutiny of AI-mediated commerce intensifies.

Amazon's Counter-Bet

Not every major brand is playing the GEO game. The most consequential strategic divergence belongs to Amazon, which has explicitly declined to list products on ChatGPT's shopping surfaces while simultaneously investing heavily in its own Alexa+ agent.

The logic is straightforward, if risky: if Amazon sends its products into a third-party agent ecosystem, it loses the customer interaction data — and the direct customer relationship — that its entire retail business model depends on. Alexa for Shopping (formerly Rufus), running on Anthropic's Haiku 4.5 model via Amazon Bedrock, is Amazon's bid to own the agent relationship rather than participate in someone else's.

The same counter-logic is available, in principle, to large retailers with sufficient brand scale and engineering resources. Bain identifies three strategic paths: embrace third-party agents (appropriate for lower-traffic retailers who benefit from exposure), build proprietary agents (viable for brands with scale), or selectively fortify the home site while opening structured data APIs. The report is explicit that "do nothing" is not a viable option on any of these paths.

The tension in this choice is real. A retailer that refuses to participate in Google AI Mode or ChatGPT shopping risks disappearing from those surfaces entirely — losing the 10%+ of revenue that AI-referred purchases represent and growing. A retailer that fully embraces third-party agents risks becoming the product supplier behind someone else's customer relationship.

There is no clean solution. The brands that navigate it best will likely be those that have made the GEO/AEO investments to remain visible in agent recommendations and have built sufficient direct-agent infrastructure to retain customers when they do convert.

Adobe's Holiday Data and the Clock

The scale of the transition isn't speculative. Adobe's 2025 holiday shopping data found AI-driven e-commerce traffic grew 758% year-over-year between November 1 and December 1, with Cyber Monday AI traffic up 670%. This year's holiday season — the first with Google AI Mode at one billion users, with Alexa+ deployed at scale, with Profound Agents running for 500+ enterprise marketing teams — will be the first real test of how much the AI citation gap matters at holiday scale.

Profound's Aim, launching now in early July, gives enterprise marketing teams roughly five months to build the operational workflows, content programs, and structured data investments that will determine their AI search visibility by November. That is not a comfortable runway. The brands that treat this summer as a preparation window rather than a watch-and-wait period will enter the holiday season with a real advantage.

The GEO investment cycle is also longer than traditional paid media. A brand that starts paying for Google Ads in September will see results in October. A brand that decides to pursue AI citation authority in September may not see meaningful shifts in AI recommendation behavior until well into 2027. The backlog of authoritative content, structured data, entity signals, and third-party coverage that LLMs weigh has been accumulating for years; moving it takes time.

What Brands Should Do Now

The thread running through all of this — Profound's $96M raise, the 57% brand-switching figure, Bain's "fulfillment pipe" warning, Amazon's decision to build its own agent rather than participate in someone else's ecosystem — is the same: in agentic commerce, visibility and control are separating. Brands that don't actively maintain AI search visibility will cede it, not to one competitor, but to whichever brands have done the structural work to earn AI citations.

The practical response has three layers, in order of time-to-impact.

Structured data first. Machine-readable product information — real-time pricing, live inventory, delivery commitments, specifications, return policies — is the fastest win because AI agents need to ingest it cleanly before they can recommend confidently. Merchants who already publish clean data feeds for Google Shopping or third-party aggregators have a head start. Those who haven't are invisible to agents making fast, specific recommendations. This investment also benefits multiple distribution channels simultaneously and ages well regardless of which AI platform wins market share.

GEO and AEO programs second, but started now. Building AI citation authority requires publishing original research, establishing entity signals, accumulating expert coverage in sources that LLMs weight as reliable, and maintaining clean structured schema. This is not a campaign — it's an ongoing program, and its effects compound over time. The brands already investing here are building a durable advantage over those treating it as a future-quarter initiative.

Measurement before all else. Without visibility into AI citations — which platforms mention your brand, how accurately, how often, and against which competitors — every other investment is directionally guessed. Profound Aim, AIVO Standard, and comparable platforms are the first generation of tools that make this visible. The specific platform matters less than the decision to start measuring. Brands flying blind on AI visibility in the second half of 2026 are not in a neutral position; they're losing ground they cannot see.

The 2026 holiday season will be the first real stress test of which brands have built AI citation infrastructure and which have not. The preparation window is five months. The signal from Lightspeed, Sequoia, and Kleiner Perkins is unambiguous: this is not a trend to monitor. It's infrastructure to build.

Sources