AI Search

AI Assistants Hit 76 Billion Monthly Sessions, Graphite Estimates

AI assistants now see an estimated 76 billion sessions a month worldwide, according to Graphite, a number so large that the company says it can no longer reliably compare AI search use with search engines. The figure marks a moment when AI assistants stop being a side channel for marketers to watch and start being a major way people get answers, with all the measurement problems that come with a shift that big.

What Graphite estimated

Graphite estimates that AI assistants reached 76 billion monthly sessions worldwide. The number covers the assistants people talk to when they want an answer, a draft, a summary, or a recommendation, rather than a list of links to click through. It is an estimate, built from the data Graphite can observe, and like any estimate of a fast-moving market it should be read as an order of magnitude rather than an exact census of every conversation.

The more interesting part of the report is what Graphite stopped doing. The company says it can no longer reliably compare AI search use with search engines, which means the old headline of AI assistants versus Google is breaking down as a measurement exercise. That is not a small admission from a company whose job is to measure this stuff, and it tells you something about how different the two behaviours have become.

Why the comparison no longer works

A search engine session and an AI assistant session are not the same unit, even though both start with a question. A search is usually a short exchange: a query, a results page, a click, done. An assistant conversation can wander across many turns, mixing research, drafting, follow-up questions, and tasks that have nothing to do with finding a web page at all. Counting one as the equal of the other flatters one side and shortchanges the other, depending on which you count.

There is also the question of what counts as a session in the first place. Different assistants define activity differently, some conversations happen inside other apps, and a single user need can be spread across a search, an assistant chat, and a document in the same hour. That suggests the honest way to talk about AI usage is to track it on its own terms, with its own metrics, rather than forcing it into a search-shaped box that no longer fits.

What 76 billion sessions means in practice

A number that large is easy to quote and hard to use. The practical question for a marketer is not how many sessions exist in the world, but how many of them touch your topic, your brand, or your product category. An assistant can have tens of billions of sessions and still rarely mention a given business, if that business never appears in the sources the assistant trusts. The headline number describes the size of the room, not your share of the conversation inside it.

That is why the measurement gap matters more than the headline. If the industry cannot agree on how to compare AI use with search, then budgets, forecasts, and client reports built on that comparison will keep wobbling. The practical takeaway is to measure what you can actually observe: how often your brand is cited in assistant answers, which sources those answers draw on, and whether that visibility turns into visits, signups, or sales.

How marketers should respond

The first response is to stop treating AI visibility as a single score to chase and start treating it as a set of behaviours to understand. Test the assistants your customers actually use, ask them the questions your customers ask, and note which brands and sources appear. That kind of manual checking is slow, but it is the only way to learn what an assistant says about your niche today, and it beats a dashboard number you cannot explain.

The second response is to invest in the material assistants cite: clear, factual, well-structured content from sources with a real reputation. Assistants lean on pages they can parse and trust, and they favour content that answers a question directly instead of burying the answer under marketing copy. That is the same work good SEO has always rewarded, now applied to a new kind of reader that happens to be a machine.

Why it matters for SEOs and link builders

For SEOs, the Graphite estimate is a signal to take AI visibility seriously as its own discipline, with its own measurement, rather than as a footnote to search traffic. The traffic an assistant sends may still be small compared with Google, but the influence of its answers on what people believe and buy is growing, and that influence will not show up cleanly in a sessions-versus-searches chart.

For link builders, the message is that citations are becoming a form of visibility in their own right. A link or a mention inside an assistant answer reaches a user who never saw a results page, and the sources that earn those mentions tend to be the ones other sites already link to and trust. Building genuine authority, through content worth citing and links worth earning, is still the most reliable route into the answers people actually see.

Sources

Sources: Search Engine Journal, November 9, 2026.

Related reading on this site: Google Tests AI Overview Query Refinements That Lead to AI Mode, 45% of Consumers Now Use AI Tools to Find Local Businesses, BrightLocal Finds, ChatGPT Ads Reach 89 Countries as 100,000 Advertisers Try the Channel, Perplexity's Decisions API Can Now Check the Web Before It Answers, Search Console Performance Report Now Lets You Filter by Multiple Devices, and Performance Max May Stop Serving Brand Searches, Google Docs Now Say.

More from Linkbaits

Turn this playbook into links

Linkbaits helps you create content that earns citations and backlinks. Start with the toolkit and build your first linkable asset this week.

Start free trial