ChatGPT Referral Traffic to B2B Sites Up 303% in 2026

Key takeaways

• ChatGPT-referred visits to B2B websites rose 303% year over year, from roughly 645,000 in June 2025 to 2.6 million in June 2026, according to Demandbase Labs.
• The growth wasn't gradual: volume more than doubled in a single month during a sharp inflection point in May 2026.
• ChatGPT is pulling ahead of other AI assistants: Perplexity referrals to the same B2B sites declined over the period, while Gemini and Claude stayed roughly flat.
• Most of this activity happens before a buyer ever lands on your site, so content built for AI citation (structured claims, clear sourcing, direct answers) now matters as much as content built for search rankings.

ChatGPT referral traffic to B2B websites grew 303% in a year, rising from about 645,000 monthly visits in June 2025 to 2.6 million in June 2026. The finding comes from Demandbase Labs, which analyzed more than 11 billion website visits across 1,584 platform instances between June 2025 and July 2026 (source: ppc.land). The growth wasn't steady: volume more than doubled during a sharp inflection point in May 2026, then held at the higher level. Over the same window, referrals from Perplexity declined, while Gemini and Claude stayed roughly flat, meaning ChatGPT is now the dominant driver of AI-assistant traffic to B2B properties. For revenue teams, the practical takeaway is this: a measurable and fast-growing share of buyer research now starts inside an AI chat window, not a search results page, and the sites and posts that get cited there are winning attention your team can't buy back later in the funnel.

What Demandbase Labs actually found

Demandbase Labs pulled the data from its own customer platform instances, tracking referral traffic patterns across more than a year of activity (source: ppc.land). The headline number: monthly ChatGPT-referred visits climbed from around 645,000 in June 2025 to 2.6 million by June 2026, a 303% increase.

That growth was not linear. The study identified a sharp inflection point in May 2026, when referral volume more than doubled compared to prior months before settling into a new, higher baseline. That kind of step-change pattern usually signals a shift in how a platform surfaces links (in this case, likely changes to how ChatGPT cites and links to sources in its answers) rather than a slow build in user habits.

The scale of the underlying dataset is worth noting too: 11 billion website visits is large enough to smooth out noise from any single company or industry, which is part of why the 303% figure is being picked up and cited across marketing trade press, including coverage from Demand Gen Report and MarTech Cube. When multiple independent outlets corroborate the same underlying number from the same study, it's a reasonable basis for a revenue team to act on, even before it shows up in your own analytics.

The study also separated ChatGPT from the rest of the AI assistant field:

• ChatGPT referrals: up 303% year over year, with an accelerating trend.
• Perplexity referrals: declined over the same period.
• Gemini and Claude referrals: stayed roughly flat.

That divergence matters for prioritization. If your team has limited bandwidth to optimize for AI-assistant discovery, this data points toward ChatGPT as the highest-volume channel to get right first.

Why this is happening now

Three forces are converging. First, ChatGPT's user base has scaled into the hundreds of millions of weekly active users, and a growing share of that usage is professional and research-oriented rather than casual. Second, ChatGPT has expanded how often and how visibly it cites external sources in its answers, which directly creates clickable referral paths that didn't exist in the same volume a year earlier. Third, B2B buyers increasingly use AI assistants as a first research step before they ever open a vendor's website, similar to how search engines became a default first stop in the 2000s.

None of this means traditional search is dying. It means a second discovery layer has opened up alongside it, and that layer runs on different rules: direct answers, clear sourcing, and content structured so a model can extract and cite a specific claim.

It also means the gap between "written for a search engine" and "written for an AI assistant" is closing, but it hasn't fully closed. A page can rank well in traditional search while still being a poor candidate for AI citation if it buries its main point under three paragraphs of scene-setting. The pages and posts that win in both worlds tend to be the ones that answer the question first, then support that answer with specifics.

What this means for revenue teams, not just SEO teams

It's tempting to file this under "marketing's problem." That's a mistake. When a buyer asks ChatGPT to compare vendors, summarize a category, or explain a trend, the answer it gives shapes the shortlist before your sales team ever gets a call on the calendar. If your company's content, and your reps' LinkedIn content, isn't structured to be cited in that answer, you're invisible at the exact moment a deal's shortlist is forming.

This is also why the split between company content and personal content matters more than ever. AI assistants pull from a mix of owned web content and public social content when forming an understanding of a company and its people. A revenue team whose founder, AEs, and CS leads are consistently publishing clear, citable insight on LinkedIn builds a second surface area for AI discovery that a static "About" page can't match on its own.

How to position your content to capture this shift

A few concrete moves compound quickly here:

1. Answer the question in the first two sentences. AI assistants extract and cite direct answers far more often than they extract buried conclusions. Lead every article and every LinkedIn post with the core point, not a preamble.
2. Add explicit stats with sources. Content that cites a real number and attributes it is more extractable and more citable than vague claims. This applies to blog content and to LinkedIn posts referencing data.
3. Structure for scanning. Short paragraphs, clear H2s, and FAQ blocks are easier for a model to parse and quote than dense prose.
4. Keep your team's LinkedIn presence active and consistent. Individual profiles carry disproportionate reach and visibility compared with company pages, and that same visibility advantage extends to how AI assistants encounter and cite your people. Social Sprint's own research has found B2B buyers already lean on AI to research vendors but don't trust it alone, which means the humans behind the brand still have to show up and back up what the AI surfaces.
5. Don't treat this as a one-time audit. Referral patterns shift with each model and product update, as the May 2026 inflection shows. Revisit your content's structure quarterly, not annually.

What to watch next

Two things are worth tracking as this trend develops. First, whether Perplexity's decline reverses: a pullback from one assistant while another accelerates suggests the AI-assistant referral market is still consolidating, not settling into a stable multi-platform split. Second, whether Gemini and Claude's flat referral numbers change as those products expand citation behavior in their own answers; a flat trend today doesn't rule out a similar inflection point tomorrow. For a broader look at how AI is reshaping the buyer's path to a vendor, see how AI is already changing the LinkedIn content game for B2B sales teams.

FAQ

Q: How much did ChatGPT referral traffic to B2B websites grow?
A: According to Demandbase Labs, ChatGPT-referred visits to B2B websites grew 303% year over year, from about 645,000 monthly visits in June 2025 to 2.6 million in June 2026.

Q: Is this growth unique to ChatGPT, or are other AI assistants driving similar traffic?
A: No. Over the same period, Perplexity referrals to the same B2B sites declined, while Gemini and Claude stayed roughly flat. ChatGPT is currently the dominant driver of AI-assistant referral traffic to B2B properties.

Q: What caused the sharp increase in May 2026?
A: The Demandbase Labs study identified a sharp inflection point in May 2026, when referral volume more than doubled compared to prior months. The report didn't attribute a single cause, but a jump of that size typically points to a change in how the platform surfaces and links to sources in its answers.

Q: Does this mean traditional SEO no longer matters?
A: No. Traditional search remains a major discovery channel. AI-assistant referral traffic is a second, fast-growing layer on top of it, not a replacement for it, and it rewards a different kind of content structure: direct answers, clear sourcing, and scannable formatting.

Q: What should a B2B revenue team do differently because of this data?
A: Structure website and LinkedIn content to answer questions directly and cite real sources, keep individual team members' LinkedIn presence active and consistent (not just the company page), and revisit content structure regularly since referral patterns can shift quickly, as the May 2026 inflection shows.

The bottom line

A 303% jump in ChatGPT referral traffic in a single year is a signal, not a footnote. B2B buyers are increasingly forming their first impression of a vendor inside an AI chat window, and the content that gets cited there, whether it lives on your website or on your team's LinkedIn profiles, is shaping shortlists before sales ever gets involved. Teams that structure their content to be directly answerable and clearly sourced now will be the ones showing up in those answers next year. Social Sprint helps revenue teams turn that individual LinkedIn presence into a consistent, trackable system: see how it works in the dashboard.