Adobe published its September retail figures on October 29. AI referred traffic to US retail sites converted 90% better than non-AI traffic, and produced 83% higher revenue per visit, per Adobe's own analysis. The data covers more than one trillion visits, 100 million SKUs and 18 product categories.
Just over a year earlier, the same dataset said the opposite. In August 2025, AI traffic converted 9% worse than non-AI traffic.
That is not a channel growing. That is a channel inverting.
So does AI search traffic convert better than everything else you are paying for? As of September, yes, and not by a little.
And it is not the story most business owners have been told about AI search. That story is about stolen clicks. On its October 29 earnings call, Reddit's CEO said Google's AI Mode does not drive traffic, the same week Reddit told shareholders its search referrals had dropped by more than 1 million a week in October. If you have been following what Google shipped this year, none of that is a surprise. The lost impressions are real. But if you spent this year defending click volume, you defended the wrong metric.
TL;DR
- Adobe's September data: AI referred visits converted 90% better and generated 83% more revenue per visit than non-AI visits. In August 2025 the same traffic converted 9% worse.
- AI traffic to US retail sites grew 71% year over year, and Adobe projects it rises 159% on Thanksgiving, 95% on Black Friday and 88% on Cyber Monday.
- Volume is still small, so this never shows up in a sessions report. It shows up in revenue per visit.
- Adobe scored the average US retail homepage at 63% on its Citation Readability Score. General merchandise homepages scored 40%.
The Number That Flipped Is the Only One Worth Reading
Almost every AI search statistic published this year measured volume. Impressions lost. Clicks lost. Zero-click rates.
Adobe measured value, and the value went the other way. In September, AI referred visits produced 83% higher revenue per visit, which Adobe called the largest gap it has observed this year. The conversion gap was 90%, the highest Adobe has recorded.
Nothing about your website caused that reversal. The assistants got better at qualifying, which is a reminder that the model was never your advantage. Someone arriving from ChatGPT or an AI Overview has already had the comparison conversation and narrowed the options. They are not researching. They are confirming.
Does AI Search Traffic Convert Better? The On-Site Behavior Explains Why
The engagement numbers rule out the easy read that these are bargain hunters bouncing off a product page. AI referred visitors spent 73% more time on retail sites, viewed 25% more pages per visit, and added products to cart at a 53% higher rate.
That is someone doing final diligence on pages you probably have not touched in two years.
Small Channel, Wrong Dashboard
Here is why almost nobody has acted on this.
AI traffic is still a sliver. In a case study shared this week, SEO consultant Cyrus Shepard opted one site out of Google's AI features and estimated AI impressions converted to clicks at roughly 0.9%. He told Search Engine Roundtable that AI impressions were about 8% of that site's total. One site, and Shepard himself called it small, so treat the figures as directional.
Directional is enough. The channel is small, high-value and growing fast. Your traffic report is built to flag the opposite: large, low-value, declining. You only see this if you read revenue per visit by source, and most businesses we audit never have. It is the same pattern as AI adoption generally, where most companies use the tools and under 9% actually deploy it.
Thanksgiving Is the Stress Test, Not Black Friday
Adobe's holiday projections are not evenly distributed, and the shape is the useful part. AI driven retail traffic is projected to rise 159% year over year on Thanksgiving, 95% on Black Friday, and 88% on Cyber Monday. Adobe also forecasts that AI's revenue per visit advantage will double this holiday season.
Thanksgiving is the biggest jump by a wide margin, which makes sense. That is the day people shop from the couch with a phone and an assistant instead of a browser and twelve open tabs. It is also the day your team is off and nobody is watching the site.
Adobe Graded Your Homepage and Nobody Sent You the Score
Buried in the same release is a metric that got almost no coverage. Adobe's Citation Readability Score measures how much of a page's content AI can understand and cite.
The average US retail homepage scored 63%, meaning more than a third of homepage content is invisible to the assistants sending shoppers. Electronics led all categories at 79%, driven by its category pages (79%) and customer service pages (78%). Grocery came in at 55%. General merchandise homepages scored 40%.
The pattern is not industry. It is whether a page explains. Electronics retailers publish setup guides, troubleshooting articles and detailed specs. General merchandise homepages are navigation-heavy and promotional, built around a carousel.
An assistant cannot cite a carousel. If your store is losing that fight, our e-commerce and conversion work starts there.
If You Do Not Sell Products, This Still Lands
The Adobe data is retail, so apply it carefully. The mechanism travels anyway. A Boca Raton law firm or a West Palm Beach HVAC company gets the same visitor: someone who already asked an assistant what this costs and how long it takes, landing to verify a fact or two before calling. If those facts live only in a PDF or a sales conversation, you just lost a visit worth nearly double. That is a question of how the funnel should be built, not of traffic.
What To Do This Week
- Segment AI referrals in GA4 and compare revenue per visit. One segment covering chatgpt.com, perplexity.ai, claude.ai, gemini.google.com and copilot.microsoft.com, measured against organic search over the last 90 days. If it outperforms, you have your budget case. If the segment is empty, your attribution is broken before your strategy is.
- Rewrite the first 40 words of your five highest-intent pages to state a fact an assistant can quote. Price range, service area, turnaround time, what is included, who it is not for. Plain sentences, not a hero headline.
- Move your change freeze and your on-call coverage to the Tuesday before Thanksgiving. Test checkout and lead forms on mobile Safari and in-app browsers specifically.
FAQ
Does AI search traffic convert better than regular search traffic?
According to Adobe's September 2026 analysis of over one trillion visits to US retail sites, yes. AI referred visits converted 90% better and generated 83% more revenue per visit than non-AI traffic. That reverses August 2025, when the same traffic converted 9% worse. The data is retail specific, so other industries may differ.
Should I block AI crawlers to protect my traffic?
The evidence argues against it for most commercial sites. A small October 2026 case study by Cyrus Shepard found that when one site opted out of Google's AI features, its AI impressions went to zero and its regular organic clicks dropped too. Blocking forfeits your highest-value traffic to defend a metric that is not paying you.
How do I make my pages easier for AI assistants to cite?
Answer one question per page in plain text near the top, include specifics such as prices, timelines and eligibility, keep key facts in HTML rather than images or PDFs, and maintain real customer service and policy pages. Adobe found electronics led every retail category because its category and support pages are full of setup guides, troubleshooting articles and specs.
The Reframe
Every AI search conversation for two years has been a defense conversation. How much traffic am I losing, what do I block, how do I get cited.
The September data makes it a yield conversation. A channel that converts 90% better does not need volume to matter. It needs a site that can close.
Book a 30-minute AI search strategy call. No pitch, just an honest read on whether your site can convert the traffic the assistants already send you. Get a free strategy call.