AI traffic converts (much) better than traditional traffic – but most websites aren't readable by AI

AI is no longer just a marketing gimmick: it's becoming the primary interface between consumers and brands, especially in e-commerce. In the US, a growing share of traffic already comes from AI interfaces (chatbots, AI browsers, personal assistants) that recommend products and redirect to merchant sites.

Adobe has just released a massive report based on more than one trillion visits to US e-commerce sites and a survey of more than 5,000 consumers. The findings are twofold:

• AI-sourced traffic is exploding.
• This traffic now converts significantly better than non-AI traffic (SEO, SEA, email…).

But there's a problem: a large share of websites simply aren't readable by AI engines – which limits their visibility in the responses of these new intermediaries.

In this article, we cover what Adobe found, why AI traffic converts better, and above all what retailers should do now to make their sites "AI-ready".

1. What the Adobe report shows: AI is already a standalone traffic channel

Adobe analyzes live transaction data from more than one trillion visits to US e-commerce sites, making it one of the largest sources of online shopping behavior data in the world. On top of that, a survey of more than 5,000 US consumers covers AI usage across the purchase journey.

1.1. AI traffic is exploding

In the first three months of 2026 (January–March), traffic coming from AI sources to US retail sites grew by nearly 400% year-over-year. In March 2026, it was still up nearly 270% compared to March 2025. During the holiday period (November–December 2025), growth was even more spectacular, with AI traffic multiplied by nearly 8x year-over-year.

In parallel, 39% of surveyed consumers say they have already used AI for online shopping, and 85% believe it improves their experience (faster search, better deals, better product relevance).

1.2. AI traffic converts better than the rest

The most striking change: AI traffic now converts far better than other channels.

• In March 2026, AI-sourced visits produced a 42% higher conversion rate than non-AI traffic (paid search, email, etc.).
• A year earlier, in March 2025, that same traffic was converting… 38% worse.

In other words, in one year, AI traffic went from "experimental traffic" to the best acquisition channel in terms of conversion.

Adobe links this shift to two phenomena:

• Rising trust: 66% of respondents now consider AI tools to deliver accurate results.
• Rapid experience improvements: better models, better integrations, better redirect journeys to sites.

1.3. Longer engagement, richer sessions

Adobe's data also shows that:

• Users arriving via AI have a 12% higher engagement rate.
• They spend 48% more time on site.
• They view 13% more pages per session.

Logical: these visitors often arrive with already-qualified intent (they asked AI a specific question, filtered products, compared options…). It's not a "curious" click, it's a click from someone who has already completed part of their purchase journey inside the AI interface.

2. The big problem: most sites aren't readable by AI

What changes everything with AI is that visitors no longer land on your site "by chance". Between your brand and your prospect sits a new filter: language models deciding which products, which pages, which content are relevant to answer a request.

And Adobe shows that large portions of e-commerce sites aren't readable by these models.

2.1. The AI Content Visibility Checker: an AI-readability thermometer

Adobe built a diagnostic tool, the AI Content Visibility Checker, which analyzes any web page and evaluates what LLMs (large language models) can and can't read. Each page gets a score out of 100%. A 50% score means half of the content is invisible to AI models.

In US retail, the averages are as follows:

• Homepages: ~75%
• Category pages (e.g. "appliances", "men's clothing"): ~74%
• Product pages: ~66%

Product pages are particularly problematic: they contain the critical information (descriptions, specs, prices…), but a large share of that content isn't structured or exposed in a way models can read.

For other page types, scores range around 73–82% (store locator, FAQ, customer service, returns, loyalty, etc.). That means 20 to 30% of content is often "off-radar" to AI.

2.2. Huge gaps between leaders and laggards

On homepages, Adobe notes a significant gap:

• Best-in-class sites: ~82.5% readability.
• Worst sites: ~54.2%.

We're talking about a gap of more than 50% between leaders and laggards. In a world where AI becomes the main entry point, this readability gap will translate into a massive visibility gap… so traffic… so revenue.

> Key message for retailers: it's not AI that "picks" a brand at random. It picks the sites it can understand, index, and summarize.

3. SEO vs AEO: why you need to think "AI readability", not just Google ranking

For 20 years, we've optimized sites for Google with classic SEO: tags, internal linking, content, speed, Core Web Vitals, etc. This work is still crucial… but no longer enough.

AI interfaces (chatbots, browsing assistants, shopping agents) don't just read the same signals as Google Search:

• They need structured, readable, semantic content.
• They need to extract clear information (price, specs, benefits, shipping terms, return policy…).
• They need to understand the relationships between pages (products, categories, FAQ, customer service, store locator…).

We're moving from SEO (Search Engine Optimization) logic to AEO (AI Engine Optimization) logic.

In practice, this means:

• Working on content clarity (less decorative text, more informative text).
• Structuring pages with explicit sections, lists, and tables.
• Using structured data (schema.org, JSON-LD).
• Making pages "queryable" by AI (content accessible, not buried in images, scripts, or opaque JS components).

4. What retailers (and more broadly all sites) should do now

From Adobe's insights, we can draw a very concrete action plan for e-commerce sites (and for many B2B sites too).

4.1. Audit AI readability on your key pages

First step: diagnose. Identify the most critical page types:

• homepage,
• categories,
• product pages,
• FAQ / support,
• store locator / points of sale,
• account / loyalty pages.

For each page type:

• Check the content's readability level.
• Analyze what's visible and invisible to AI: text, descriptions, specs, FAQ, micro-content.

Goal: get an "AI score" per page template and a shortlist of quick wins.

4.2. Rewrite and restructure content for the models

Then rework high-traffic, high-value pages.

Product pages

• More explicit descriptions.
• Structured spec sheets (size, color, materials, compatibility…).
• Clear user benefits (use cases).
• Embedded product FAQ.

Category pages

• Introductory text explaining what's in this category.
• Usage context (for whom, for what, how to choose).
• Internal links to product families and best-sellers.

FAQ / support pages

• Questions and answers in simple wording.
• Vocabulary close to how consumers actually ask AI their questions.

The idea isn't to produce more text for the sake of producing more text, but to make every page understandable without context, for a machine that lacks human "intuition".

4.3. Make your data machine-friendly

Beyond editorial content, site data needs to be actually usable:

• Systematic use of structured data (product, reviews, FAQ, organization, store, offer, etc.).
• Avoid key info being stuck only in images, PDFs, or iframes.
• Make sure frontend components don't hide info from bots (lazy loading, client-only JS, late-injected content, etc.).

The goal: enable an AI model to accurately answer questions like:

• "Show me men's running shoes, size 42, with cushioning, delivered within 48h."
• "Which of this retailer's stores are open within 10 km of me, today?"
• "Which are the best beginner models in this category?"

5. Why it's an opportunity (not just a risk)

From a distance, this Adobe report can be scary: "If my site isn't readable by AI, I'm going to disappear from new traffic channels."

But it's also a massive opportunity for those who move fast:

• AI traffic is already here.
• It's growing much faster than other sources.
• It converts much better than traditional traffic.
• Most sites aren't ready: there's still little competition on this ground.

Business translation:

• The first brands to become AI-ready will capture a disproportionate share of this new qualified traffic.
• They'll be over-represented in AI assistant, AI browser, and conversational shopping interface responses.
• They'll improve conversion without scaling acquisition budgets as much.

6. How Busony helps you become AI-ready

At Busony, we help e-commerce merchants and brands move from a "SEO-only" site to a fully AI-ready site:

• AI readability audit of your key pages (home, categories, products, FAQ, store locator…).
• Targeted rewrite of content and structure to improve your "AI score".
• Structured data implementation and AEO (AI Engine Optimization) best practices.
• Support for integrating AI into your overall traffic and conversion strategy.

Discover our Agentic Optimization offering — also available in français.