How Much Traffic Do AI Assistants Send Compared With Google?
A low single-digit share in both traffic studies. Haynes found that ChatGPT sessions were 0.62% of the volume Google organic search drove across 100 sites, and 0.73% with all other LLM tools included (Marie Haynes). The SEO Works found that AI assistants sent 2.6% of the traffic organic search sent, or 72,095 sessions against 2,825,594 (The SEO Works).
| Measure | Marie Haynes | The SEO Works |
|---|---|---|
| Sites | 100 GA4 properties, over 145 million sessions | 95 GA4 properties; 63 used for conversion comparisons |
| Period | 12 months (end month not stated) | September 2025 to August 2026 |
| AI traffic relative to organic | ChatGPT 0.62%; all LLMs 0.73% (of Google organic) | 2.6% (organic delivered 39 times more) |
| Conversion rate, AI visitors | ChatGPT 13.8% | 7.18% |
| Conversion rate, organic visitors | 3.02% | 6.08% |
| Conversion volume | Google organic drove 27 times more conversions than all LLMs combined | 5,001 AI conversions against 155,605 organic (3.2 per 100) |
The SEO Works defines AI referrals by referring domain, matching sources such as chatgpt.com and claude.ai with a regular expression, because GA4’s own “AI Assistant” channel only exists from June 2026. Its eligibility rule for conversion comparisons was at least 100 AI referral sessions and 1,000 organic sessions in the period (The SEO Works).
Is AI Referral Traffic Growing?
The two studies point in different directions. Haynes writes that in August 2025, “in conjunction with the release of GPT-5, the exponential growth of ChatGPT sessions ended,” and that ChatGPT “has been sending fewer and fewer visitors to websites since then” across her 100 properties (Marie Haynes). She publishes that trend as a chart without the underlying monthly numbers.
The SEO Works reports a rise from May 2026: AI referrals moved from 2.29% of organic sessions between September 2025 and April 2026 to 3.04% from May onward, reaching 3.86% in August. It links the timing to ChatGPT making brand names clickable inline around May 7, 2026, but says it treats that connection “as a working hypothesis rather than an established cause,” and that four months is too short a window for firm conclusions (The SEO Works).
AI sessions as % of organic (Haynes: % of Google organic; The SEO Works: % of organic search)
Source: Marie Haynes, "GEO is a distraction," October 21, 2026; The SEO Works, "AI Referral Traffic Statistics Benchmark" (monthly table). Read October 22, 2026.
The two studies use different sites, periods and AI-source filters, so compare rows within a study, not across studies. Haynes does not state the end month of her 12 months. Neither study can count Google AI Overviews clicks, which arrive as ordinary organic sessions.
In every row, AI assistants sent less than 4% as many sessions as organic search, and the highest single month in The SEO Works data came after ChatGPT began making brand names clickable in May 2026.
Do AI Referral Visitors Convert Better?
On average, yes, but by very different margins and not on every site. Haynes found ChatGPT visitors converted at more than 4.5 times the rate of Google organic visitors, 13.8% against 3.02%, on the 55 sites tracking conversions. The SEO Works found a much smaller gap, 7.18% against 6.08%, and AI traffic converted worse than organic on 23 of its 63 websites; five of those recorded no AI conversions at all from 1,487 AI sessions (The SEO Works).
The comparison channel changes the answer. The same SEO Works data gives AI referrals a conversion ratio of 0.67x against paid search, 1.18x against organic, 1.37x against total sessions and 2.48x against direct traffic. By customer type, AI converted at 1.51x organic for B2C sites, 1.46x for mixed, 1.21x for B2B services and 0.62x for B2B products (The SEO Works). The agency warns that “Anyone quoting a conversion multiple for AI traffic without stating the comparison channel has published a number that cannot be checked or reproduced.”
Does AI Visibility Lift Direct and Branded Traffic?
Neither study that tested it could show that it does. Indig looked for AI visits hidden in GA4’s Direct channel across 12 sites and found that Direct sessions with identifiable AI traces were only 0.003% of total Direct traffic on average. Much of the Direct growth he saw was bots: at one site, GA4 recorded 409,000 Direct sessions, of which only 198 counted as engaged (Growth Memo).
His separate test of four brands found apparent correlations of 0.72 and 0.47 between AI visibility measures and Direct traffic, but when he deliberately misaligned the weeks, “several wrong alignments produced equally strong or stronger correlations.” His conclusion: “An effect could still exist. I just couldn’t isolate it” (Growth Memo).
The SEO Works ran a similar check on what it calls the “billboard” theory, that people see a brand in an AI answer and later type its address. After controlling for overall site growth, the correlation between AI growth and direct growth was −0.230 (p = 0.075), and it found no relationship with direct conversions. What it did find was that AI referral growth and organic growth moved together, with a controlled correlation of +0.534 (The SEO Works).
Why Might the Studies Undercount AI Influence?
Because referral analytics only count clicks that carry an identifiable source. The SEO Works says Google AI Overviews clicks “arrive as ordinary organic sessions and cannot be separated in GA4,” so “the measured AI contribution should be read as a floor.” It also lists in-app browsers, stripped referrers and people who read an AI answer and then search for the brand by name as visits its method would miss (The SEO Works).
Indig cites one counter-example he did not test himself: on the Siege Media Podcast, growth expert Eric Wu said Ramp identified 30% to 40% of its direct traffic as coming from answer engines, using page-level patterns and self-reported attribution (Growth Memo). We have not seen Ramp’s data. For the measurement side, see our guides on how much traffic AI assistants refer and tracking AI referral traffic in GA4.
What Did the Authors Conclude, and What Did the Headline Add?
The authors concluded less than the headline did. Haynes argues that “chasing ChatGPT citations and LLM mentions is optimizing for the wrong target” and that the AI search actually sending people to businesses is Google’s own, through AI Overviews and AI Mode. The SEO Works calls AI referral traffic “real but small” and says becoming stronger in AI depends heavily on SEO visibility. Indig recommends combining where a brand appears in AI answers, how customers say they found it, and commercial results (Marie Haynes; The SEO Works; Growth Memo).
Search Engine Roundtable’s headline, “AI Visibility & Search Not Worth Investing In,” is the outlet’s summary (Search Engine Roundtable). None of the three says AI answers have no influence; each says the influence is either small in clicks or not yet measurable. All three authors sell search services or paid subscriptions, and Indig’s issue carried a sponsored segment from Semrush.
What This Means for Anyone Buying Coverage
Do not budget coverage on the promise of AI referral traffic, and do not dismiss AI mentions because the sessions are few. On this evidence, measured clicks from AI assistants are a low single-digit share of organic traffic, so a placement should be justified by the readers, credibility and search visibility it delivers directly. The SEO Works finding that AI growth tracks organic growth suggests the same groundwork, consistent information and genuine authority, serves both.
A paid article or sponsored post is advertising: it is labeled, disclosed, and its links are typically nofollow or sponsored. Nothing in these studies shows that payment buys AI citations or AI traffic. Our recommendation, not a rule: set up conversion (key event) tracking before judging any channel, since 45 of Haynes’s 100 sites tracked none, and ask any vendor quoting an AI conversion multiple for the sample, period and comparison channel behind it.
Methodology and Limitations
Four limits apply. First, the samples are small and anonymous: Haynes says her 100 sites were “randomly sampled” without describing how, and The SEO Works replaces site names with index numbers. Second, the studies are not comparable with each other, because their periods, denominators and conversion definitions differ; Haynes compares with Google organic, The SEO Works with all organic search. Third, we read the published pages, not the underlying data; The SEO Works offers its aggregated dataset for download, which we did not test. Fourth, GA4 counts key events, not people, so The SEO Works describes every conversion figure as “an upper bound on the number of real enquiries.”
Conclusion
AI assistants are a real but small referral source in these three analyses, and their visitors often, though not always, convert well. Whether being named in AI answers builds demand that analytics cannot see remains unproven in both studies that tested it. Buy coverage for what it does directly, and measure before you pay for visibility alone.
Frequently Asked Questions
How much traffic does ChatGPT send compared with Google?
In Marie Haynes’s study of 100 websites, published October 21, 2026, ChatGPT sessions were 0.62% of the volume Google organic search drove, or 0.73% with other LLM tools included.
What share of website traffic comes from AI assistants?
The SEO Works found AI assistants sent 2.6% as many sessions as organic search across 95 websites from September 2025 to August 2026, ranging from 2.01% in February 2026 to 3.86% in August 2026.
Do visitors from AI assistants convert better than search visitors?
Often. Haynes reports 13.8% for ChatGPT against 3.02% for Google organic, and The SEO Works reports 7.18% against 6.08%, but AI converted worse than organic on 23 of The SEO Works’ 63 sites.
Does appearing in AI answers increase direct or branded traffic?
Not measurably in these studies. Kevin Indig found no reliable relationship across four brands, and The SEO Works found no evidence that AI growth drives direct traffic once overall site growth is controlled for.
Are Google AI Overviews counted as AI traffic?
No. The SEO Works says AI Overviews clicks arrive as ordinary organic sessions in GA4 and cannot be separated, so its AI share should be read as a floor.
Can a paid placement buy AI citations or AI traffic?
Nothing in these studies shows that. A paid placement is advertising, labeled as such, and no placement guarantees AI citations, traffic or rankings.
Sources and Methodology
This Data Report is built from three analyses (Haynes’s dated October 21, 2026, Indig’s October 19, 2026, and The SEO Works’ undated page covering data to August 2026) and one trade-press roundup, all read on October 22, 2026. Every figure is copied from the authors’ own pages; the conversion-volume and trend statements are theirs. Conclusions about what buyers should do are our recommendation, not a rule. The SEO Works page carries no publication date. We did not re-run any analysis or open the underlying data, and we did not review the other studies The SEO Works cites for comparison.
- vendor research Marie Haynes, GEO is a distraction, October 21, 2026. Sample, ChatGPT and LLM traffic shares, conversion rates, 96.5% conversion share, 45 sites without key events, trend claim. Read October 22, 2026.
- vendor research The SEO Works, How Much Traffic Do AI Assistants Actually Send, And Does It Convert Better? (AI Referral Traffic Statistics Benchmark), data September 2025 to August 2026. Sessions, monthly shares, conversion ratios, segments, correlations, methodology and limitations. Read October 22, 2026.
- vendor research Kevin Indig, Growth Memo, Does AI drive more direct traffic?, October 19, 2026. Direct-traffic and correlation tests, Ramp example as relayed. Read October 22, 2026.
- trade press Search Engine Roundtable, 3 New Studies Suggest AI Visibility & Search Not Worth Investing In, October 22, 2026. Identifies the three studies and the headline framing. Read October 22, 2026.