The AI Visibility Gap: Why Brands Are Missing 45 Billion Monthly Conversations

TL;DR: New research from Graphite reveals that AI tools now generate 45 billion monthly sessions worldwide - equivalent to 56% of global search engine volume. The industry has been underestimating AI usage by 4-5x because 83% of it happens in mobile apps, invisible to standard web analytics. Google's traffic share dropped from 89% to 71% in just two years. Meanwhile, our analysis of nearly 1 million AI prompts shows that 92-99% of AI citations come from topic-specific sources most brands aren't optimizing for. The result? A massive AI Visibility Gap - billions of conversations where your customers are asking AI for recommendations, and your brand doesn't exist. This report breaks down the data, explains why it matters, and shows what to do about it.

Table of Contents

1. The 4-5x undercount: AI is much bigger than anyone measured

2. Google's shrinking share: 89% → 71% in two years

3. The AI Visibility Gap: what brands can't see is hurting them

4. What our data shows: how AI actually picks sources

5. The commerce trigger: AI will buy, not just recommend

6. The 92% blind spot: where brands should actually be

7. Closing the gap: a data-driven framework

8. FAQ

The 4-5x undercount: AI is much bigger than anyone measured

For the past two years, the industry narrative has been that AI search is interesting but small - maybe 5% of Google's volume, growing but not yet material. That narrative is wrong by a factor of 5.

Research published by Graphite CEO Ethan Smith analyzed AI usage across the five largest LLM products - ChatGPT, Gemini, Perplexity, Grok, and Claude - and compared them with the six largest search engines. The findings are striking:

45 billion monthly AI sessions worldwide. That's not a typo. Forty-five billion. For context, that equals 56% of global search engine volume and 34% of search volume in the United States, where AI tools receive 5.4 billion monthly sessions.

Why has everyone been undercounting? Because 83% of all AI usage happens in mobile apps (75% in the US). Standard web analytics tools - SimilarWeb, Semrush, GA4 - measure desktop browser sessions. They miss the vast majority of AI usage entirely. As Graphite's analysis demonstrates, most previous comparisons between AI use and search engine use underestimated AI by 4-5x.

This isn't a niche technology anymore. The total combined usage across search engines and AI tools has grown 26% since 2023. The pie is expanding, not just shifting. But the distribution of that pie has changed dramatically.

The platform breakdown

ChatGPT dominates. According to Graphite's data, it drives 89% of global LLM usage and commands roughly 20% of worldwide search share when adjusting for prompts that are actual queries. In the US, ChatGPT holds approximately 12% of search traffic share.

But ChatGPT is not alone. Today, 810 million people use ChatGPT daily. Google AI Overviews reaches 1.5 billion users monthly. Perplexity has crossed 45 million monthly active users. These platforms aren't fighting over a small segment of early adopters - they are where mainstream users go for answers.

Google's shrinking share: 89% → 71% in two years

Graphite's data paints a clear picture of market shift. Worldwide, Google's traffic share has decreased from 89% in 2023 to 71% in Q4 2025. In the US, the decline went from 88% to 75% over the same period.

Let that sink in. Google lost 18 percentage points of global search share in two years. Not because people stopped searching - total search and AI usage grew 26% - but because a massive volume of queries migrated to AI platforms.

What this means for marketing

For 20 years, "search optimization" meant "Google optimization." Brands built entire marketing teams, tooling stacks, and reporting frameworks around one platform. Google Analytics. Google Search Console. Google Ads. Google Merchant Center.

That playbook is no longer sufficient. If Google now represents 71% of global search rather than 89%, brands that only optimize for Google are structurally missing up to 29% of the discovery landscape. And within Google itself, the AI transformation is accelerating: Google AI Mode generates 93% zero-click searches. Standard AI Overviews produce 83% zero-click results. Even traditional Google results without AI end in zero clicks 60% of the time.

The user behavior shift is clear: people are getting answers without clicking. If your brand isn't mentioned in those answers - in both Google's AI features and independent AI platforms - you're invisible during the majority of discovery interactions.

The AI Visibility Gap: what brands can't see is hurting them

Here's the problem that Graphite's data exposes at scale: brands have a sophisticated analytics infrastructure for the 71% (Google), and essentially zero visibility into the other 29% (AI platforms).

Think about what most brands can measure today:

• Google organic: rank position, impressions, clicks, CTR (Search Console)

• Google paid: impressions, clicks, conversions, ROAS (Google Ads)

• Social: reach, engagement, followers (native dashboards)

• Direct/referral: sessions, conversions (GA4)

Now think about what most brands cannot measure:

• Does ChatGPT recommend them when users ask for products in their category?

• What does Perplexity say about their brand vs. competitors?

• How often is their brand cited across 45 billion monthly AI sessions?

• What sentiment does AI attach to their brand when mentioning it?

• Which competitor is AI recommending instead?

This is the AI Visibility Gap: the delta between where your customers are discovering products (increasingly in AI) and where your brand is visible and measurable (mostly in Google and social). According to research spanning 80 million+ citations across 10,000+ domains, 67% of information discovery now happens through AI interfaces. Two-thirds. Most brands are blind to most of it.

The hidden cost

The AI Visibility Gap doesn't show up in any traditional dashboard. Google Analytics won't show you the traffic you didn't get. Search Console won't show you the mention that didn't happen. Your monthly marketing report looks fine - organic traffic is stable, paid is performing, social is growing. But underneath, AI is steering thousands of potential customers to competitors, and you have no signal.

AI referral traffic currently accounts for 1.08% of all web traffic and is growing by approximately 1 percentage point per month - making it the fastest-growing traffic source in internet history. ChatGPT alone drives 87.4% of all AI referral traffic. And this traffic converts: AI referrals convert at 4.4x to 23x the rate of standard organic traffic, because users arrive with the trust of an AI recommendation already in place.

If you're not measuring your AI visibility, you're not measuring where your market is going.

What our data shows: how AI actually picks sources

Knowing that AI search is massive is step one. Step two is understanding how AI decides which brands to mention - and which to ignore. This is where GetMentioned's proprietary data adds a critical layer.

Our analysis of nearly 1 million prompts across all major AI models reveals the mechanics of AI citation behavior.

AI overwhelmingly favors topic-specific sources

This is the single most important finding for any brand trying to close the AI Visibility Gap. When ChatGPT or Perplexity generates a recommendation, 92% of the sources it draws from are niche, topic-specific domains - industry publications, product review sites, expert blogs, comparison pages, technical documentation. Only 8% comes from general platforms like Wikipedia, Reddit, or LinkedIn.

Gemini is even more extreme: 99% topic-specific, leaving just 1% for general sources.

The citation mechanics

Our data reveals several patterns that determine whether AI mentions your brand:

The front-loading effect: 44.2% of all AI citations reference content from the first 30% of a page. AI models scan content and disproportionately weight what appears early. If your brand name, key differentiators, and data points are buried in the middle of your articles, AI is less likely to pick them up.

The 4-slot competition: ChatGPT typically cites 4 unique sources per answer. That means for any given user query, your brand is competing against the entire internet for one of four citation slots. Making it into those slots requires being in the right sources, with the right content structure, at the right level of authority.

The search trigger window: Only 31% of ChatGPT prompts trigger a web search - the rest are answered from training data. And within a conversation, the first question is 2.5x more likely to trigger citations than the 10th question, and nearly 4x more likely than the 20th. This means the window for your brand to be cited via real-time search is narrow - you need to already be in training data OR be highly visible in search results for AI's RAG (Retrieval-Augmented Generation) to find you.

The cross-verification pattern: AI models don't just find a single source and cite it. They look for confirmation across multiple independent sources. If your brand appears consistently on Reddit, in industry publications, on review sites, and on your own website - all with the same positioning and messaging - AI gains confidence to recommend you. Inconsistent or contradictory information across sources reduces your citation probability.

The platform divergence: Only 11% of domains are cited by both ChatGPT and Perplexity. The platforms have dramatically different source preferences. Optimizing for one doesn't guarantee visibility in the other. A multi-platform strategy is essential.

Which sources matter most?

• Wikipedia: 7.8% of all ChatGPT citations - the single most-cited general domain

• Reddit: 6.6% of Perplexity citations, 2.2% of Google AI Overviews - critical for Perplexity visibility

• .com domains: Over 80% of all ChatGPT citations - commercial and industry sites dominate

• .org domains: 11.29% of ChatGPT citations - non-profits and organizations carry weight

If your brand's online presence is limited to LinkedIn posts and a Wikipedia mention, you're operating in roughly 8% of the citation space that AI draws from. The other 92% - the industry publications, the expert roundups, the product comparisons, the technical blogs - that's where AI is actually building its recommendations.

The commerce trigger: AI will buy, not just recommend

Graphite's data establishes the scale of AI search. Our data explains the mechanics. But there's a third dimension that makes the AI Visibility Gap even more consequential: AI is evolving from recommendation engine to purchasing agent.

Google launched the Universal Commerce Protocol (UCP) on January 11, 2026 at the NRF conference, co-developed with Shopify, Walmart, Target, Etsy, Wayfair, and endorsed by 20+ partners including Visa, Mastercard, and Stripe. UCP is an open standard that enables AI assistants - ChatGPT, Gemini, Perplexity - to complete full purchases on behalf of users within the AI conversation.

The numbers back the urgency:

• AI platforms are projected to drive $20.9 billion in retail spending in 2026 - nearly 4x the 2025 figure

• Over 50% of consumers are predicted to use AI shopping assistants by end of 2026

• ChatGPT already enables US users to buy from 1M+ Shopify merchants within the conversation

• Cart abandonment (currently ~70%) drops dramatically when checkout happens inside AI - no redirects, no forms, no friction

The funnel collapse

Before UCP:

User asks AI → AI recommends brand → User leaves AI → Navigates to site → Maybe buys

After UCP:

User asks AI → AI recommends brand → AI completes purchase → Done

The entire discovery-to-purchase funnel collapses into a single AI conversation. This transforms the AI Visibility Gap from a brand awareness problem into a direct revenue problem. If AI doesn't recommend your brand, the user doesn't just see a competitor - they buy from a competitor, right there in the conversation.

The cost of being invisible in AI just went from "lost awareness" to "lost sales."

The 92% blind spot: where brands should actually be 

Combining Graphite's scale data with our citation analysis reveals a compounding blind spot.

Layer 1 (Graphite): 83% of AI usage happens in mobile apps, invisible to web analytics. Brands don't even know the volume of AI conversations happening about their category.

Layer 2 (GetMentioned): 92-99% of AI citations come from topic-specific sources that most brands aren't actively optimizing for. Even brands that understand AI search tend to focus on Reddit and Wikipedia - which represent only ~8% of the citation space.

Layer 3 (Platform divergence): Only 11% of domains are cited by both ChatGPT and Perplexity. Brands that optimize for one platform miss the other.

Layer 4 (Commerce): With UCP, invisibility in AI now directly costs revenue, not just awareness.

Stack these layers and the picture becomes clear: most brands are invisible in most AI conversations, on most AI platforms, across most of the sources that AI actually uses, while those conversations increasingly end in purchases.

The six factors that determine your AI visibility

Our analysis of millions of AI citations identifies six weighted factors:

Citation Frequency at ~35% is the dominant signal. It's not enough to be mentioned once on one authoritative site. AI needs to see your brand consistently across multiple credible, topic-specific sources to build the confidence to recommend you.

Closing the gap: a data-driven framework 

The AI Visibility Gap is wide, but it's closeable. Here's a framework grounded in the data:

Step 1: Measure what you can't see

You can't close a gap you can't measure. Start with a baseline:

• Run a free AI Visibility Report to see what ChatGPT, Perplexity, and Gemini say about your brand today

• Identify 30-50 key queries your customers would ask AI about your category

• Check your brand's presence (or absence) for each query across platforms

• Map competitor visibility on the same queries

This gives you your AI Visibility Score - your starting point.

Step 2: Fix the source gap

Remember: 92-99% of AI citations come from topic-specific sources. Audit your brand's presence across the sources that actually matter:

• Industry publications and media

• Product review sites (G2, Capterra, Trustpilot)

• Expert blogs and comparison articles

• Technical documentation and how-to guides

• Industry-specific directories and associations

If your brand is absent from these sources, AI has no raw material to build a recommendation from. Getting mentioned in 5-10 authoritative topic-specific sources for your category is the single highest-leverage action.

Step 3: Front-load your content

44.2% of citations come from the first 30% of content. Restructure your key pages:

• Brand name and core value proposition in the first paragraph

• Key data points and differentiators above the fold

• Direct answers to common questions early in the article

• Don't bury your competitive advantage in section 7 of an 8-section page

Step 4: Build cross-platform consistency

AI cross-verifies across sources. Ensure your brand messaging, positioning, and key claims are consistent across your website, industry mentions, Reddit discussions, review sites, and thought leadership content. Inconsistency reduces AI confidence; consistency increases it.

Step 5: Publish data, not just opinions

Content with original data - statistics, benchmarks, research findings - has 30-40% higher visibility in AI responses. AI models weight factual, data-rich content over opinion pieces. Every major piece of content should include 5-10 specific, sourced statistics.

Step 6: Optimize for freshness

AI prefers current sources. Add "last updated" dates to all key content. Refresh quarterly with new data, updated statistics, and current examples. An article dated March 2026 significantly outperforms the same content dated 2024 in AI citation probability.

Step 7: Monitor continuously

AI responses are non-deterministic - the same query can produce different answers hour to hour. One-time audits give you a snapshot, not a trend. GetMentioned provides continuous monitoring across ChatGPT, Perplexity, Gemini, and other platforms, tracking your Share of Voice, citation frequency, sentiment, and competitive positioning over time.

FAQ 

How big is AI search compared to Google? According to Graphite's research, AI tools generate 45 billion monthly sessions globally - equivalent to 56% of search engine volume worldwide and 34% in the US. Google's global traffic share has decreased from 89% in 2023 to 71% in Q4 2025. The total market (search + AI) has grown 26% since 2023.

Why has AI usage been undercounted? Because 83% of all AI usage happens in mobile apps (75% in the US), which standard web analytics tools don't measure. Graphite's analysis is one of the first to combine desktop web sessions with mobile app data, revealing that previous estimates underestimated AI usage by 4-5x.

What is the AI Visibility Gap? The AI Visibility Gap is the delta between where consumers are discovering products (increasingly through AI platforms - 45B monthly sessions) and where brands are visible and measurable (mostly Google and social). 67% of information discovery now happens through AI interfaces, but most brands have zero analytics coverage for these interactions.

How does AI decide which brands to recommend? Our analysis of nearly 1 million prompts shows AI overwhelmingly favors topic-specific sources (92-99% of citations) over general platforms like Wikipedia or Reddit (~1-8%). The dominant factor is Citation Frequency (~35% weight) - how often your brand appears across credible, topic-specific sources. Other factors include Position Prominence (~20%), Domain Authority (~15%), Content Freshness (~15%), Structured Data (~10%), and Technical Performance (~5-10%).

How can I check if AI recommends my brand? Start with a free AI Visibility Report - enter your brand name and see what ChatGPT, Perplexity, and Gemini say about you in seconds. For continuous monitoring, GetMentioned tracks AI mentions, Share of Voice, sentiment, and competitive positioning across all major AI platforms.

Why does this matter for revenue, not just awareness? Google's Universal Commerce Protocol (UCP), launched January 2026, enables AI to complete full purchases inside the conversation. AI retail spending is projected to reach $20.9 billion in 2026, nearly 4x the 2025 figure. When AI can buy on behalf of users, invisibility in AI doesn't just cost awareness - it costs direct sales.