Get in Touch
GEO & AI Search

Claude Is the #2 AI Engine Nobody Optimizes For

Updated 5 min read Daniel Shashko
Claude Is the #2 AI Engine Nobody Optimizes For
AI Summary
Multiple independent datasets in mid-2026 show that Claude has become the number two AI engine for business audiences. Goodie's B2B panel found Claude rose from 1.4 percent to 18.5 percent of AI referral traffic in eight months, passing Gemini and Perplexity into second place. Previsible found Claude grew 64 times over and overtook Perplexity in March 2026, though across the whole web ChatGPT still sends 92.4 percent of AI referrals. First Page Sage puts Claude second by usage at 21.5 percent, and Claude ranks second on the iOS App Store. The catch is that Claude is a blind spot: it shows no consistent citation panel, sometimes answers from training data with no sources at all, and AI-visibility trackers added Claude support later than the other engines. OrganikPI ran the same two buying questions used in its four-engine test through Claude, confirming it answered the CRM question from memory with zero citations. The advice for brands is to check GA4 for claude.ai referrals, optimize for entity strength and the roundups Claude cites when it does search, and choose a tracker that genuinely covers Claude.

Almost every brand putting effort into AI search is optimizing for ChatGPT. While they did, a second engine passed Gemini and Perplexity and became the number two source of AI referral traffic for business brands. It is Claude, and most companies cannot see it in any dashboard they own.

Several independent datasets landed this quarter, and they point the same way. Goodie’s AI Search Traffic Report tracked a panel of B2B brand sites and found Claude went from 1.4 percent of their AI referral traffic to 18.5 percent in eight months, moving into second place ahead of Gemini and Perplexity. Previsible’s referral study found Claude grew 64 times over and overtook Perplexity in March 2026. First Page Sage’s usage market share report puts Claude at 21.5 percent, also second. Even on the iOS App Store, Claude now ranks second in both downloads and revenue, ahead of Gemini. This is the shift we track across AI search, and it is why we keep saying a single-engine plan misses buyers.

How big is Claude, really

It depends on what you count, and being precise here matters. Measured across the entire web, ChatGPT still dominates referral volume at 92.4 percent in Previsible’s data, and Claude is a small slice of that. Measured on B2B brand sites, Claude is second at 18.5 percent. Measured by raw app usage, it is second at 21.5 percent. The one result that holds across every dataset is that Claude passed Perplexity, and for business audiences it passed Gemini too. So measured across all web traffic, ChatGPT is still most of the story. Measured on B2B brand sites, Claude is very likely your second-largest AI referral source right now, and you are probably not measuring it.

Two bar charts. The first shows share of B2B AI referral traffic, with ChatGPT at 62.6 percent, Claude second at 18.5 percent, Gemini at 10.6 percent and Perplexity at 7.3 percent. The second shows Claude rising from 1.4 percent in mid-2025 to 18.5 percent in early 2026.
On B2B brand sites, Claude is now the second-largest AI referral source, up from almost nothing eight months earlier. Source: Goodie AI Search Traffic Report 2026.

Finding 1: you cannot see it, because Claude barely shows its work

Claude becomes a blind spot the moment you try to measure it. We asked it the same two buying questions we recently put to four other engines in our four engines, four shortlists test: the best CRM and the best email marketing platform for a small business. The answers were useful. What they did not include is the thing every AI-visibility tool depends on.

For the CRM question, Claude did not search the web at all. It answered from training data, labelled its own step “Synthesized established CRM knowledge”, and cited zero sources. There is no citation panel to appear in and no ranked list of pages that won. For the email question it did search and it did cite, but the sources it showed were Insider, Brevo, Salesforce and Mailercloud Blog, mostly roundups and vendors’ own blogs rather than independent pages.

Claude answering the best CRM platforms for a small business, recommending HubSpot, Pipedrive, Zoho, Salesforce, Freshsales, Monday and Airtable, with a status line reading synthesized established knowledge and no cited sources shown.
We asked Claude for the best CRM platforms. It answered from training data, labelled the step ‘Synthesized established CRM knowledge’, and cited no sources at all.

The inconsistency is what makes it hard. Because Claude retrieves and cites on one query and answers from memory on the next, a tracker can log that claude.ai sent you a visitor without telling you which answer or which source produced the visit. The measurement surface the other engines give you is mostly absent here.

Claude answering the best email marketing platforms for a small business, recommending MailerLite, Moosend, Brevo, HubSpot, Klaviyo, Omnisend, Mailchimp and Flodesk, with inline source labels such as Insider, Brevo and Salesforce.
On the email question Claude did search and cite, but the sources were competitor blogs and roundups rather than the vendors’ own pages.

Finding 2: the trackers were late to Claude

The tooling reflects the same lag. Most AI-visibility platforms built their coverage around ChatGPT, Perplexity, Gemini and Google first, because those engines show sources you can scrape. Claude support arrived later across the category. Profound, one of the larger trackers, only added support for the Claude Fable model on June 9. We have reviewed several of these tools hands-on, including Otterly.AI, AthenaHQ and Peec AI, and Claude depth is exactly where they differ most. If your tool does not cover Claude well, your second engine is missing from your reports.

What Claude actually recommends

Claude’s shortlists reinforce what we found across the other engines. For CRM it named the same safe core that showed up everywhere, HubSpot, Pipedrive and Zoho, then added Airtable, a pick no other engine made. For email it led with MailerLite, the way Gemini and Perplexity did, then named Flodesk, which again no other engine surfaced. Claude is effectively a fifth distinct shortlist. It agrees on the household names and diverges in the tail, the same pattern we found when we compared Google’s two AI surfaces. Each engine has to be earned on its own, Claude included.

What this means for your brand

  • Look for claude.ai in your referral report today. It is the one Claude number you can see without any tool. Open GA4, filter referrals, and check the trend. For a B2B brand it may already sit second behind chatgpt.com.
  • Optimize for memory and for the sources Claude does pull. Because Claude often answers from training data, being an established, consistent entity across the web matters more than ranking one page. When it does search, it favors roundups and comparison pages, so earning a place in those is how you show up. This is the same source and entity work behind the GEO tools we recommend.
  • Pick a tracker that genuinely covers Claude. Coverage varies more than the marketing suggests. Our hands-on look at the trackers that rival Profound and our roundup of the ones we rate are the fastest way to see which ones measure it and how deeply.

The bottom line

Claude just became the second biggest AI engine for business audiences, and most brands have not measured it once. It sits outside most dashboards, answers partly from memory, and shows few sources when it does answer. That combination makes Claude the hardest engine to influence and the easiest to leave out of a report. Start by looking for claude.ai in your traffic, then decide how much of your AI visibility work has been aimed at a single engine. If you want a picture of who every engine names today, our free audit of where the engines name you and our tracker roundup are the place to begin.