AI commerce conversion rate: what the data actually says

The most-quoted conversion number in agentic commerce is wrong on three counts, and this site helped spread it. Here is the correction, the figures Adobe actually published, and the two multi-store datasets that measure absolute ecommerce conversion rates.

AI-referred visits to US retail sites converted 60% higher than non-AI visits in July 2026, the eleventh straight month ahead (Adobe Analytics via Digital Commerce 360, 19 August 2026). In absolute terms, ChatGPT referrals converted at 1.81% across 94 ecommerce stores in 2025 (180 Marketing, 2026).

Last updated: 13 October 2026.

Correction: where the 15.9% vs 1.76% stat really comes from

From May 2026 until this update, this page, our homepage stat bar and several other posts on this site said that ChatGPT-referred ecommerce traffic converts at 15.9% versus 1.76% for Google organic, credited to Adobe (2025). Other publications picked the line up from us. On 10 October 2026 we traced the number to its primary source. Adobe did not publish it, in 2025 or at any other time.

The pair belongs to Seer Interactive, a digital marketing agency in Philadelphia. Its case study "6 Learnings, 1 Site: How Traffic from ChatGPT Converts", by Nick Haigler and Garman Chan, published 3 June 2025, reports GA4 key-event conversion rates for one Seer client between 1 October 2024 and 30 April 2025: "Google Organic - 1.76%. ChatGPT - 15.9%. Perplexity - 10.5%. Claude - 5% Gemini - 3%" (Seer Interactive, 2025).

Three facts about that study decide how it can be used:

• One site. The authors write, "This is just one client's data, but it's part of a bigger shift." The client's industry is not named (Seer Interactive, 2025).
• Leads, not purchases. The conversions are GA4 key events on a lead-generation funnel: 1,370 AI-driven conversions worth "over $205,000 in the pipeline", framed as whether visitors take "meaningful action, like signing up for a service or requesting a demo". The words ecommerce, retail, purchase and revenue do not appear (Seer Interactive, 2025).
• A tiny sample. AI traffic was "just under 11,000 sessions versus almost 14 million" for Google organic over seven months, well under a tenth of a percent of the organic volume (Seer Interactive, 2025).

How Adobe's name got attached is mundane. In our operating notes the Seer line sat next to a figure that is Adobe's, the 4,700% year-on-year growth in AI-referred traffic to US retail sites in July 2025 (Adobe, August 2025), and the attribution bled from one line to the next. The word "ecommerce" was never in Seer's text at all. So the 15.9% figure is Seer Interactive's, describes lead conversions on one B2B site, and cannot be used in an ecommerce sentence or as a benchmark for anyone. This page is corrected; our other pages that carried the claim are being updated to the figures below.

What Adobe actually measured, month by month

Adobe Analytics reports relative conversion: how AI-referred visits to the US retail sites in its dataset convert compared with visits from every other source. It has never published an absolute "ChatGPT converts at X%" figure. The arc is the real story.

Adobe's own summary: "A year ago, AI-referred visitors converted at nearly half the rate of non-AI traffic. In March 2026, they converted 42% more", on more than one trillion visits to US retail sites and more than 100 million SKUs (Adobe Digital Insights, April 2026).

For most of 2025, the period the misattributed figure supposedly described, Adobe found AI-referred retail traffic converting worse than other traffic, not nine times better. The premium that exists now is tens of percent, less than a year old, and growing.

Absolute conversion rates: what the ecommerce datasets show

A relative lift needs an anchor. Two multi-site studies provide one.

180 Marketing, under its Visibility Labs banner: 94 stores. Jeff Oxford's team pulled GA4 data for 94 ecommerce stores across January to December 2025: 9.46 million non-branded organic sessions against 135,000 ChatGPT sessions. ChatGPT referrals converted at 1.81% versus 1.39% for non-branded organic, a 31% lift, on an average order value of $204 against $238, 14.3% lower (180 Marketing, 23 February 2026). The authors note that shoppers often take a ChatGPT recommendation to Google and search the brand name, so 1.81% is likely an undercount (180 Marketing, 2026; Search Engine Land, 2026).

Kaiser and Schulze, Marketing Science: 973 sites. The largest academic sample covers 973 ecommerce sites with about $20 billion in revenue, more than 50,000 ChatGPT-referred transactions against 164 million from traditional channels. Organic LLM traffic was under 0.2% of visits, and in its first year it showed "conversion rates and revenue per session above paid social but below all other traditional channels", improving month by month (Kaiser and Schulze, Marketing Science, 2026).

The three sources agree: AI-referred traffic is a small share of visits, converts better than non-branded organic in the 2026 Adobe and 180 Marketing data, trailed most channels in the academic sample's first year, and carried a smaller basket where order value was measured. A ninefold advantage appears in none of them. Before any AI conversion number goes into a deck, check who published it, whether the conversion is a lead or a purchase, whether the rate is absolute or relative, how many sites and sessions sit behind it, and which months it covers. A case study is published because its result is striking; a benchmark is published on a schedule whatever it shows.

What changes for a brand

Stop budgeting on a borrowed number. The spread between a single-site lead-gen case and a 94-store purchase dataset is an order of magnitude. Your own AI-referred conversion rate is the only one that matters, and GA4 can report it once assistant referrers are grouped into an AI channel; our ChatGPT Ads in GA4 guide covers the setup.

Count the visits that arrive as Direct. Most assistant clicks carry no referrer. Scot Wingo of ReFiBuy calls this Dark Agentic Commerce Traffic; citing Loamly's 446,405 tracked visits, he puts the share of AI-influenced visits arriving with no referrer at 70.6% (Retailgentic, 7 April 2026). Our DACT guide covers the mechanics. Whatever rate GA4 shows for your AI channel is computed on a fraction of the real sessions.

Reconcile against a second signal. For MyMuse, GA4 reported ₹11.5K/mo of AI-referred revenue. A 12-month attribution model reconciling GA4 sessions against post-purchase survey and checkout source signals put real AI revenue at ₹81.2K/mo, 7× the GA4 reading, and surfaced +₹29.34L of Year 1 net additional revenue (Tru Commerce case study, 2026).

Separate paid from organic AI. ChatGPT Ads clicks arrive tagged; organic assistant referrals mostly do not, and blending them produces a rate that describes neither. Our ChatGPT Ads benchmarks hold the paid-side numbers, including the roughly 1.3% average CTR in our own US campaigns since July 2026 (our campaign data).

Decide where the sale completes. Adobe's premium is measured on visits that reach a retailer's site; as checkout moves inside the assistant, the click stops being the unit of measurement. Our map of AI shopping assistants shows which surfaces complete the purchase where; the argument is in from recommended to transacted.

Where Tru Commerce fits

We sell the transaction side of this: ChatGPT Ads campaigns, an on-site AI shopping assistant and agentic checkout. On measurement, MyMuse is the reference: GA4 showed ₹11.5K/mo of AI revenue, the real figure was ₹81.2K/mo, 7× hidden, and the Year 1 net surfaced was +₹29.34L. If you want your AI-referred conversion rate measured rather than borrowed, book a demo.