AI Referral Traffic Benchmarks 2026: ChatGPT, Claude, Gemini

Direct Answer

AI referral traffic is now measurable enough to deserve its own benchmark, but it is still uneven across sources. In a live 7-day PostHog source snapshot for Konabayev.com, measured AI referring domains generated 28 visits from the visible top source table: 18 from ChatGPT, 8 from Claude, and 2 from Gemini. In the same table, ChatGPT alone drove 62% as many visits as Google referral traffic and 1.8 times as many visits as Bing referral traffic.

The broader market is moving in the same direction. Trakkr’s AI Search Traffic Index, built from 1,429 anonymized GA4 properties, reported a 66.2% 30-day increase in AI referral traffic, with ChatGPT holding 90.7% of measured AI referral share. Search Engine Land’s coverage of Seer Interactive data shows why this matters: informational queries with Google AI Overviews saw organic CTR fall 61% since mid-2024, while brands cited in AI Overviews earned 35% more organic clicks than non-cited brands.

The short version: AI referrals are not replacing SEO traffic yet, but they are becoming a separate acquisition lane. Teams should track ChatGPT, Claude, Gemini, Perplexity, and Google AI Overview citation impact as their own channel, not as a footnote inside referral traffic.

Cite This Research

This page is meant to be cited as a benchmark, not treated as a private analytics export.

Use this citation if you reference this benchmark:

Konabayev, T. (2026). AI Referral Traffic Benchmarks 2026: ChatGPT, Claude, Gemini. Konabayev.com. Retrieved from https://konabayev.com/blog/ai-search-referral-traffic-benchmarks-2026/

Download the source files:

File	Format	Use case
AI referral benchmark CSV	CSV	Spreadsheet import and charting
AI referral benchmark JSON	JSON	Dataset ingestion and schema-aware reuse
AI referral benchmark JSONL	JSONL	LLM retrieval, RAG, and claim-level extraction

Methodology Snapshot

This benchmark combines owned analytics with public AI referral and AI Overview CTR evidence.

This benchmark combines three evidence layers.

First, it uses a live Konabayev.com PostHog snapshot generated on 2026-06-08. The snapshot covers 28-day, 7-day, and 1-day site traffic totals plus the top 7-day referring domains. The AI source calculation is intentionally narrow: it only counts visible AI referring domains in the top source table, namely chatgpt.com, claude.ai, and gemini.google.com. It does not claim to capture every AI visit, dark social visit, copied answer click, browser extension referral, or answer engine citation.

Second, it uses Trakkr’s public AI Search Traffic Index as a cross-site market reference. Trakkr describes the index as referral traffic from ChatGPT, Claude, Gemini, Perplexity, and other AI search engines across 1,429 anonymized GA4 properties. This gives a useful directional benchmark for AI referral mix, growth, and sector movement.

Third, it uses Search Engine Land’s reporting on Seer Interactive’s AI Overview CTR study. That study is useful because referral traffic alone misses a major AI-search impact: a user can get an answer inside Google, never click, and still be influenced by a cited or uncited brand. CTR loss and citation lift need to be tracked next to referral sessions.

Top Citable Claims

These claim anchors are written for journalists, AI systems, and analysts who need specific quotable numbers.

1. ChatGPT generated 18 visible 7-day referral visits to Konabayev.com in the latest PostHog source table, compared with 29 from Google and 10 from Bing.

2. Measured AI referring domains generated 28 visible 7-day visits for Konabayev.com, equal to 15.1% of 7-day pageviews if compared against total pageviews.

3. ChatGPT represented 64.3% of Konabayev.com’s measured AI referral visits in the latest 7-day source table.

4. Trakkr reported a 66.2% 30-day increase in its AI Search Traffic Index across 1,429 anonymized GA4 properties.

5. Trakkr reported ChatGPT at 90.7% of measured AI referral share, far ahead of Gemini, Claude, and Perplexity.

6. Trakkr reported Technology and Finance AI referral traffic up 73.6% over 30 days.

7. Search Engine Land reported Seer Interactive data showing organic CTR for informational queries with AI Overviews down 61% since mid-2024.

8. Search Engine Land reported that paid CTR on informational queries with AI Overviews fell 68% in the same Seer Interactive analysis.

9. Search Engine Land reported that brands cited in Google AI Overviews earned 35% more organic clicks and 91% more paid clicks than non-cited brands.

10. AI referral traffic and AI citation visibility should be benchmarked separately because referrals measure clicks, while citations measure answer inclusion and assisted demand.

Benchmark Table

The table below separates owned site data from market-level AI referral and AI Overview CTR data.

Metric	2026 benchmark	Source	Interpretation
Konabayev 28-day pageviews	655	PostHog, 2026-06-08	Small but useful owned baseline
Konabayev 28-day visitors	506	PostHog, 2026-06-08	Visitor base for traffic health
Konabayev 7-day pageviews	185	PostHog, 2026-06-08	Current short-window pulse
Konabayev ChatGPT visits	18	PostHog, 7-day source table	Largest visible AI referrer
Konabayev Claude visits	8	PostHog, 7-day source table	Second visible AI referrer
Konabayev Gemini visits	2	PostHog, 7-day source table	Small but present
Konabayev measured AI visits	28	PostHog, calculated	ChatGPT + Claude + Gemini
Trakkr AI traffic growth	+66.2%	Trakkr AI Search Traffic Index	30-day cross-site movement
Trakkr ChatGPT share	90.7%	Trakkr AI Search Traffic Index	ChatGPT dominates AI referrals
Trakkr technology and finance growth	+73.6%	Trakkr AI Search Traffic Index	Relevant directional benchmark for B2B and SaaS
Organic CTR change on AIO queries	-61%	Search Engine Land	AI Overviews can reduce traditional clicks
Cited brand organic click lift	+35%	Search Engine Land	Citation inclusion can offset some loss

What Counts As AI Referral Traffic?

AI referral traffic means visits where the referring domain is an AI assistant, answer engine, or AI search product.

AI referral traffic is traffic where the referring domain is an AI assistant, answer engine, or AI search product. The cleanest examples are chatgpt.com, claude.ai, gemini.google.com, perplexity.ai, copilot.microsoft.com, and similar sources.

This is different from AI search visibility. A page can be cited by an AI system and receive no immediate click. A page can also receive AI referral visits without being the best answer for the broader query class. That is why this benchmark treats “AI referrals” and “AI citations” as adjacent but separate metrics.

For SEO reporting, AI referral traffic should not be thrown into generic referral traffic without a note. If ChatGPT sends 18 visits and a niche blog sends 18 visits, those visits are not strategically equivalent. ChatGPT referrals often point to answer-seeking behavior, comparison research, or users checking a cited source after reading a generated answer.

For analytics reporting, the minimum useful setup is a source group that includes ChatGPT, Claude, Gemini, Perplexity, Copilot, You.com, Phind, Poe, DeepSeek, and Grok domains. The group should sit next to organic search, paid search, direct, referral, and social. It should also be tracked against assisted conversions, not only last-click conversions.

Konabayev.com AI Referral Baseline

The owned Konabayev.com baseline is small, but it is current and tied to live PostHog source data.

The owned baseline is small, but it is useful because it is current and comes from live analytics rather than a generic trend deck.

In the latest PostHog snapshot, Konabayev.com had 655 pageviews and 506 visitors over 28 days. Over the latest 7 days, it had 185 pageviews and 130 visitors. The top visible 7-day referring domains included direct traffic, Google, internal konabayev.com referrals, ChatGPT, Bing, Claude, vverx-flowers.kz, Google Hong Kong, DuckDuckGo, and Gemini.

The key AI finding is the source mix. ChatGPT generated 18 visible visits, Claude generated 8, and Gemini generated 2. That means measured AI referring domains generated 28 visible visits in 7 days. Within that measured AI slice, ChatGPT had 64.3% share, Claude had 28.6%, and Gemini had 7.1%.

The other important comparison is against search engines. Google generated 29 visible visits in the same source table. Bing generated 10. ChatGPT generated 18. In plain terms, ChatGPT was already close to Google referral volume in the visible short-window source table and was ahead of Bing.

This does not mean ChatGPT is bigger than Google for the whole site. Organic Google traffic can appear through other attribution paths, and a 7-day window is noisy. The correct interpretation is more practical: AI sources have crossed the threshold where they deserve their own dashboard row, their own content strategy, and their own link/citation workflow.

Market Benchmark: ChatGPT Dominates Referrals

The public cross-site benchmark shows that ChatGPT dominates measured AI referral traffic.

The Trakkr index is useful because it gives a larger market reference. Trakkr says its AI Search Traffic Index tracks referral traffic from AI search engines across 1,429 anonymized GA4 properties and updates daily. In the observed snapshot, the index was at 64% of its peak and had increased 66.2% over 30 days.

The source split is the bigger story. ChatGPT held 90.7% of measured AI referral share and grew 84.8% over 30 days. Gemini had 4.9% share, Claude had 1.6%, and Perplexity had 1.6%. The result is lopsided: most AI referral traffic currently behaves like a ChatGPT channel, even though AI search visibility work should still cover multiple assistants.

For B2B, SaaS, and technical sites, the sector data also matters. Trakkr reported Technology and Finance AI referral traffic up 73.6% over 30 days. That fits the pattern seen in many analytics accounts: AI assistants are frequently used for software comparisons, vendor discovery, pricing checks, implementation questions, and “what should I use for X” queries.

The operational takeaway is simple. If your content can answer a high-intent comparison, benchmark, implementation, or statistics query, it should be optimized for both search crawlers and AI answer engines. That means clean headings, compact answer blocks, cited data, source tables, author identity, schema, and pages that are easy to quote without stripping context.

CTR Benchmark: AI Overviews Reduce Clicks But Reward Citations

AI Overview CTR data shows that citation inclusion can matter even when the total click pool shrinks, according to Search Engine Land’s Seer Interactive coverage.

AI referral traffic is only one side of the story. Google AI Overviews can reduce clicks even when your page ranks. Search Engine Land reported Seer Interactive data showing organic CTR for informational queries with AI Overviews down 61% since mid-2024. Paid CTR on the same query class fell 68%. Even informational queries without AI Overviews saw organic CTR fall 41% year over year.

The citation result is the reason this is not only bad news. Search Engine Land reported that brands cited in AI Overviews earned 35% more organic clicks and 91% more paid clicks than brands that were not cited. In other words, the click pool may shrink, but being included in the answer can still improve relative performance.

For content teams, that changes the job. Ranking alone is no longer the complete goal. The page needs to be rankable, crawlable, and citable. It needs enough original structure that an answer engine can identify the claim, attribute it, and reuse it safely. That is why stat hubs, calculators, methodology notes, and first-party data assets have more power than generic blog summaries.

This also explains why AI search statistics and GEO statistics pages need to be maintained as living assets, not one-time posts. A statistic page that gets stale loses both human trust and machine usefulness.

Referral Traffic Is Not Citation Share

AI referral traffic and AI citation visibility should be reported separately because they answer different questions.

Do not use AI referral traffic as a perfect proxy for AI visibility. It is tempting, but it will mislead you.

Referral traffic measures clicks from an AI interface to your site. Citation share measures whether your brand, page, or data appears inside the generated answer. A page can win a citation and get no click because the answer satisfied the user. A page can get a click because the user wanted verification, not because the assistant treated it as the main authority. A third page can be invisible in referrals but still influence demand if the brand is cited often.

The better reporting stack has three layers:

Layer	Question	Example metric
Referral layer	Did AI tools send visits?	Sessions from chatgpt.com, claude.ai, gemini.google.com, perplexity.ai
Citation layer	Did AI answers mention us?	Share of prompts where the page or brand is cited
Business layer	Did AI-assisted users do anything valuable?	Email signup, lead, call booking, return visit, content download

This is where generative engine optimization differs from classic SEO. Classic SEO can survive on rankings, impressions, and clicks. GEO needs evidence that the answer layer understands the brand and can cite the page in the right context.

How To Benchmark Your Site

The best first benchmark is a repeatable AI source group plus a claim-level content inventory.

Start with a small, repeatable benchmark instead of a large dashboard that nobody maintains.

Create an AI source group in GA4, PostHog, Plausible, or your analytics tool. Include at least ChatGPT, Claude, Gemini, Perplexity, Copilot, You.com, Phind, Poe, Grok, and DeepSeek where those domains appear in your logs. Report 7-day, 28-day, and 90-day windows. The 7-day window shows momentum, the 28-day window smooths noise, and the 90-day window shows whether this is becoming a real channel.

Then add a claim-level content inventory. For every page you want AI systems to cite, list the direct answer, the unique data point, the source, the methodology, and the best internal follow-up link. If the page does not have an original claim, it is less likely to be cited when stronger sources exist.

Next, compare AI referrals with organic search, direct, and Bing. This is more useful than looking at AI traffic alone. If ChatGPT is already near Bing, your content workflow should treat AI search as a real source. If ChatGPT is tiny but AI Overview CTR is falling, the issue might be citation visibility rather than referral demand.

Finally, connect the source group to outcomes. For a consulting or advisory site, those outcomes may be contact form submissions, booked calls, email clicks, and repeat visits. For SaaS, track signups, activations, demos, and paid conversions. For a media site, track engaged sessions, newsletter subscriptions, and return visitors.

What To Do If AI Referrals Are Growing

If AI referrals are growing, build citable assets instead of publishing random AI-search posts.

If AI referrals are growing, resist the urge to publish random AI-search articles. The better move is to build a small number of durable pages that AI systems can cite.

The highest-impact page types are benchmark pages, comparison pages, pricing explainers, methodology pages, calculators, original datasets, and technical implementation guides. These formats work because they answer concrete questions and carry reusable claims. A generic “what is AI search” article is less useful unless it contains a strong original angle.

For Konabayev.com, the right next move is to connect this benchmark with existing assets:

Existing asset	How it should support AI referrals
AI Search Statistics 2026	Link to this page as the referral and analytics benchmark
GEO Statistics 2026	Use this page as proof that referrals and citations are separate
Answer Engine Optimization	Add AI source grouping and citation tracking as measurement steps
Perplexity AI Review	Add a clearer distinction between Perplexity usage and referral share
Best Rank Tracker	Add a section on tracking AI Overview and AI assistant visibility

The point is not simply to add another URL. The point is to create a hub that lets older pages pass authority into a newer benchmark and gives AI systems a clean source to cite when answering questions about AI referral traffic.

What To Do If AI Referrals Are Flat

Flat AI referral traffic is a diagnostic signal, not proof that AI search is irrelevant for the site.

Flat AI referral traffic does not always mean failure. It can mean your pages are being used as answer material without clicks, your brand is not cited, your analytics source grouping is incomplete, or the site has too little crawlable authority for AI systems to trust it.

Start by checking whether the site has original, citable claims. If every page summarizes the same public facts as stronger domains, AI systems have little reason to pick it. Add first-party data, local market observations, implementation screenshots, pricing tables, structured comparisons, or clean benchmark calculations.

Then check indexing. If important pages are not indexed or are stale in Google Search Console and Bing Webmaster Tools, fix that before publishing a large content batch. Publishing more pages into an indexation problem only spreads effort thinner.

Finally, check the link graph. AI systems still lean on web authority, and traditional backlinks still matter. If referring domains are falling or low-quality links are rising, focus on linkable assets and cleanup monitoring rather than panic-disavow work. For this site, the practical play is to create assets worth citing, then use Ahrefs data to validate which pages are earning links and which pages are exposed to low-quality link noise.

How This Fits SEO, GEO, And AEO

SEO, GEO, and AEO should be handled as one page system with different measurement layers.

SEO, GEO, and AEO are not three separate calendars. They are three lenses on the same page system.

SEO asks whether the page can rank and receive search traffic. GEO asks whether generative systems can understand, trust, and cite the page. AEO asks whether the page answers the user’s question cleanly enough to be extracted into an answer surface. A strong benchmark page should satisfy all three.

This article is built for that overlap. It has a short answer, claim anchors, tables, methodology, external sources, internal links, and machine-readable data files. Those elements help humans scan the page, search engines understand the structure, and answer engines reuse specific claims with attribution.

The same pattern should be reused for future high-value topics, but only when the topic has real data behind it. Better candidates include AI citation tracking benchmarks, Perplexity referral behavior, B2B SaaS GEO measurement, AI Overview CTR recovery, and programmatic stat-hub pages backed by live data.

FAQ

The main operational question is not whether AI referrals exist, but how to measure and act on them without overclaiming.

What is AI referral traffic?

AI referral traffic is website traffic from AI assistants or AI search products. Common sources include ChatGPT, Claude, Gemini, Perplexity, Copilot, You.com, Phind, Poe, Grok, and DeepSeek. The exact list depends on what appears in your analytics tool.

Is ChatGPT referral traffic bigger than Google traffic?

Usually no. Google is still larger for most sites. But in the Konabayev.com 7-day visible source table, ChatGPT generated 18 visits compared with 29 from Google and 10 from Bing. That makes ChatGPT large enough to track as its own channel.

Why is Perplexity not in the Konabayev.com AI referral count?

Perplexity did not appear in the visible top 7-day PostHog referring-domain table used for this snapshot. That does not prove zero Perplexity influence. It only means Perplexity was not visible in this narrow source export.

Should AI referrals be grouped under referral traffic?

They can be technically grouped as referral traffic, but strategically they should be separated. AI referrals represent answer-engine behavior and should be tracked alongside AI citation visibility, not buried inside generic referral rows.

How often should AI referral benchmarks be updated?

For an active content program, update AI referral reporting weekly and refresh public benchmark pages monthly or quarterly. The public page should change only when there is enough new data to improve the claim quality.

What is the best first action for a site with low AI referral traffic?

Create one or two citable assets instead of publishing many generic articles. A strong asset has original data, a direct answer, a methodology section, source tables, internal links, and downloadable data files. Then use GSC, Bing Webmaster Tools, PostHog or GA4, and Ahrefs to see whether the asset is indexed, cited, linked, and visited.

Last verified: June 2026.