How Much Traffic Do ChatGPT & Perplexity Send B2B SaaS Sites In 2026?

ChatGPT and Perplexity send B2B SaaS sites very little traffic today. ChatGPT referrals are about 0.33 percent of measured B2B visits, and all large language models combined deliver a median of roughly 3 percent of organic search visits. That is real growth from a small base, and not yet a reason to move budget away from organic search.

The figures apply to a typical B2B SaaS site measured by visits or sessions in mid-2026. Your own ratio can sit well above or below the median.

Every published number comes from a panel of sites with its own denominator and tracking gaps. AI clicks that arrive without a referrer are missing from all of them.

Usage Statistics Are Not Referral Traffic

Most pages ranking for ChatGPT usage statistics answer a question your finance team never asked. They count people inside a product. They do not count visits that reach your website.

ChatGPT usage statistics measure how many people use the product, while a referral session measures a click from ChatGPT that lands on your site. Only the referral session is revenue-relevant to a SaaS team.

The usual usage figures fall into four groups, and none of them tells you how much traffic to expect.

• Active users: The weekly or monthly count of people who open the product. It says nothing about how many of them click out to a website.
• Country split: Where those users are based. It describes market reach and tells you nothing about where your visits come from.
• Demographics: The age and gender mix of users. That helps a consumer brand and does very little for a B2B pipeline forecast.
• Revenue and plans: What the product earns and how many people pay for it. A strong revenue line can sit next to a tiny referral line.

This article leaves those numbers out on purpose. It answers one question. How many sessions do ChatGPT and Perplexity send to a B2B SaaS site, and is that enough to change a budget? If you need the wider definition first, read about what AI search visibility means. For adoption figures, see our AI marketing statistics for 2026.

How Much Traffic ChatGPT Sends To B2B SaaS Sites

ChatGPT is the biggest AI referrer to B2B sites, and it is still small. Demandbase Labs measured 1,584 platform instances and more than 11 billion visits between June 2025 and June 2026. ChatGPT supplied about 0.33 percent of that B2B traffic, and its referrals grew about 303 percent year on year. Growth that fast from a base that small still leaves the channel under half of one percent.

The same Demandbase Labs sample shows monthly ChatGPT referral visits rising from about 645,000 to about 2.6 million over those twelve months. Those are visits across more than a thousand platform instances, not visits to one site. A single B2B SaaS site will see a tiny slice of that total.

A second panel puts ChatGPT in context. Across 34 B2B SaaS brands in Q3 2026, ChatGPT accounted for about 89 percent of LLM referral sessions. On a like-for-like set of brands in that panel, ChatGPT sessions rose about 3.1 times between Q1 and Q3 2026.

Statcounter measures something broader. Its data shows ChatGPT at 76.85 percent of AI chatbot referrals to websites in April 2026, the lowest on record and down from 84.21 percent in April 2025. That sample covers more than 3 billion page views a month across more than 1 million sites. The share is falling, so other chatbots are taking a larger part of chatbot referrals.

The table sets the figures side by side so you can see why they do not match.

The 89 percent and the 76.85 percent are both shares of chatbot referrals, yet they differ because the sites differ. Neither is a share of your total traffic. Only the Demandbase Labs figure is.

ChatGPT Vs Perplexity: Who Sends What

ChatGPT sends the bulk of LLM referral sessions to B2B SaaS sites, and Perplexity sits well behind it. In the TripleDart panel of 34 B2B SaaS brands, ChatGPT supplied about 89 percent of LLM referral sessions in Q3 2026. Perplexity, Gemini, Claude, Copilot and every other source share the remaining 11 percent or so.

The Perplexity trend is where you should slow down. Perplexity referral sessions fell about 62 percent between Q1 and Q3 2026 in the TripleDart panel. The authors suspect a tracking change rather than a real collapse in clicks, and they advise waiting for Q4 data before reading it as a trend.

So treat the drop as a reason to track Perplexity on its own line, not as a verdict on the platform. No clean Perplexity-only share of B2B SaaS sessions turned up in the research behind this article, and we are not going to invent one. If Perplexity matters to your buyers, a dedicated look at tracking Perplexity visibility will tell you more than any panel average.

AI Referral Traffic As A Share Of Organic Sessions

For a SaaS team, the denominator that matters is organic search. The TripleDart benchmark covers 64 B2B SaaS and tech companies in June and July 2026. The median LLM visit count was about 3 percent of organic search visits. The 25th percentile sat near 1.8 percent and the 75th percentile near 4.8 percent.

That gives you a band, not a single number. Half of the companies sit between roughly 1.8 and 4.8 percent. If your ratio lands inside it, you look like a normal B2B SaaS site.

An older reference exists, and it should be read with care. An Ahrefs study of 3,000 sites, published in February 2025, found AI chatbots supplied about 0.17 percent of website visitors on average, with the top sites near 6 percent. It counts visitors, not sessions, and it predates the growth described above. Use it to see how far the channel has moved, not as a current benchmark.

Visits, visitors, sessions and traffic share are four different bases. A study that counts visitors will always read lower than one that counts sessions, because one visitor can start many sessions. Comparing a visitor-based share with a session-based share tells you nothing, and mixing denominators is the most common way these numbers get misquoted.

For the organic side of your own ratio, our guide to what happened to SEO after AI search gives the context, and the B2B SaaS content benchmarks give the organic baseline that sits under the ratio.

Why Every Measured Number Is An Undercount

Every AI referral figure, including the ones above and the ones in your own Google Analytics 4 account, is a floor. AI referral clicks are undercounted when they arrive with no referrer. GA4 then files them as direct traffic, or it shows chatgpt.com with not-set values.

The common causes are these.

• New-tab opens: A reader opens a cited link in a new tab and the referrer is dropped.
• In-app browsers: The link opens inside another app that does not pass the source along.
• The mobile app: Clicks from the ChatGPT mobile app often arrive stripped of their source.
• Plugin integrations: Links followed through browser plugins and integrations can lose the referrer in the same way.

None of this can be fixed on your side. It means the AI share in GA4 is a lower bound, and a rising direct channel can hide real AI influence. Put the same caveat on your own baseline before you show it to anyone.

How To Isolate AI Referral Traffic In GA4

Filter session source with a regex that matches the AI platform names, then build a channel group so the result stays in your reports. It works with platform names alone.

1. Write the pattern. Use the platform names you care about, ChatGPT, Perplexity, Gemini, Claude and Copilot, joined with a pipe character, for example chatgpt|perplexity|gemini|claude|copilot. Do not guess exact referrer domains. A pattern built from names is looser, which is the safer error here. You should now have one plain-text pattern to reuse.
2. Filter session source. Open a traffic acquisition view or an exploration and filter session source by that pattern. Check the list of sources that match. You should see the platforms you named and nothing unrelated.
3. Build a custom channel group. Create a channel called AI with the same session source rule and place it above Referral and Direct so it catches the traffic first. A new channel group takes 24 to 48 hours to appear in acquisition reports, so do not judge it on day one.
4. Compare against organic. Pull AI sessions and organic search sessions for the same window, ideally the last 90 days. Divide AI by organic. The result is the ratio you will test in the next section.
5. Save it for reporting. If your team reports in Looker Studio, connect the channel group there so the ratio updates without anyone rebuilding it. You should now have one number that refreshes itself.

Remember the previous section. This ratio is a floor, so a low result may be hiding some traffic in the direct channel.

The Materiality Test: When AI Referral Traffic Earns A Budget Line

This section is our reasoning, built on the published bands. It is a decision rule and not a research finding. The thresholds of 1.8 and 4.8 percent are our reading of the TripleDart quartiles.

The metric is AI referral sessions as a percentage of organic sessions over a 90-day window. Apply the rule in this order.

1. If the ratio is below 1.8 percent, track it and do not create a budget line. Your AI share sits under the 25th percentile of the TripleDart panel, so the channel is not yet material for your site.
2. If the ratio is between 1.8 and 4.8 percent, treat it as normal for a B2B SaaS site and fund measurement. Pay for a baseline and a dashboard so you can see movement quarter on quarter.
3. If the ratio is above 4.8 percent, or AI sessions convert at a rate comparable to organic, fund AI visibility work. Either signal alone is enough, because a small channel that converts well can matter more than its share suggests.

The second test uses your own numbers, because no verified conversion benchmark was found for AI referral traffic. Compare the conversion rate of AI sessions with the conversion rate of organic sessions over the same 90 days. If the AI rate is close to or above the organic rate, the third line applies even when the ratio is low.

Here is the arithmetic with a made-up round example. These figures are an illustration and not data.

In this example the ratio alone says fund measurement. The conversion test says the traffic is as valuable as organic, so the third line applies. That is why you run both tests.

The organic side is the weak point of most ratios. Semrush is a practical way to pull organic traffic and position data for the denominator, because a ratio is only as honest as its denominator and a competitor's organic trend gives you something to measure your own against. For the wider set of numbers to watch beside referral sessions, see AI search visibility metrics and KPIs. If you would rather hand over the setup, ongoing SEO and AI visibility reporting covers it.

What To Do With The Result

There are three outcomes, and each one has a trigger metric and a next step. Track when the ratio is below 1.8 percent. Build a baseline when it is between 1.8 and 4.8 percent. Fund citation work when it is above 4.8 percent or when AI sessions convert like organic.

If you land on baseline, the tools in our guide to ChatGPT answer monitoring tools are a sensible place to start. If you land on fund, read how to get cited in AI answers first, and our answer engine optimization work is the done-for-you version.

Two things are left out of this article on purpose. There is no conversion benchmark, because none was verified. There are also no tactics for getting cited, because those need their own guide.