You can see AI referrals in your analytics. However, the session count is small, and you only have an industry benchmark to tell you what an AI visitor was worth on someone else’s site.
While AI traffic volume tells you how many visitors arrived through that channel, AI traffic quality tells you whether those visitors became customers or qualified leads, and how much business each visit generated.
When AI-referred visitors convert and spend more than other visitors, your data supports more AI visibility investment. The next step is weighing those returns against the cost of attracting more AI visitors.
This guide shows you how to both assess AI traffic quality using your own analytics and connect Semrush visibility data to results after the click. We’ll also explain where Microsoft Clarity’s scrape-to-referral ratio fits into your assessment.
What does AI traffic quality mean?
AI traffic quality describes how effectively visitors from AI-generated answers contribute to your business goals. Quality is assessed through the purchases or qualified leads those visitors generate, which is then compared to what visitors from other channels produce.
Each stage of AI discovery tells you something different:
- AI visibility: How often is your brand mentioned or your content cited in the AI answers you track?
- AI referral volume: How many identifiable visits do AI platforms send to your website?
- AI traffic quality: How effectively do AI visits produce conversions and business outcomes?
Your definition of quality depends on what you count as success. For example, an online store’s traffic quality shows up in purchases and revenue per visit. For a B2B company, it shows up in leads that both match the company’s target customers and move toward a sale.
A brand that's frequently cited in AI-generated answers might see little impact on its leads. Meanwhile, a different brand with fewer citations could receive referrals that generate profitable sales — quality is what separates the two.
Why do industry studies report different results for AI traffic quality?
AI traffic quality studies reach different conclusions because they measure different audiences, conversion events, baselines, and time periods. Some industry research finds a clear conversion advantage; other studies find that AI referrals perform comparably to organic search, or even generate less revenue per visit than organic search traffic.
Semrush’s impact of AI search study reported that AI visitors were worth 4.4 times as much as traditional organic search visitors, based on conversion rates across digital marketing and SEO topics.
Siege Media’s 2026 analysis found that the median AI-to-organic conversion rate was 1.26x. Their analysis used first-party GA4 data from January through May 2026, with conversions measured using GA4’s session key-event rate.
The result varies by product
According to Shopify's Q2 2026 analysis, AI-referred shoppers were approximately 80% more likely to convert than organic visitors, but the exact difference in conversion rate varied dramatically by product. For example, the average advantage of AI-referred apparel conversion was 60%, while watches converted at 2.4 times the organic rate.
Shopify connects better results in research-intensive categories to what shoppers do before clicking. AI conversations help people compare specifications and confirm whether a product fits their needs. The shopper arrives on the site much further into the process of making the purchase decision.
The category pattern repeats across industries. Adobe’s July 2026 data showed AI-referred retail visitors converting 60% better than non-AI-referred visitors. In travel, however, AI referrals converted 1% worse. A retailer and a travel company reading the same report for their industry would therefore find different benchmarks.
But as Kyle Risley, Senior SEO Lead at Shopify, says, “Category-level analysis gives directional insights, but it doesn’t mean some are ‘good for AI’ and others aren’t.”
Conversion rate and revenue per visit can move separately
A completed conversion tells you that someone took an action, but the financial value of that conversion depends on what happened next.
Siege Media found that AI traffic matched or exceeded organic conversion rates on 72% of the sites studied. Yet among sites with reliable revenue tracking, AI generated higher revenue per session on fewer than half. Their conversion measure used GA4 key events, so the comparison also depends on which actions each business tracked.
Siege Media argues for building visibility in AI answers before the channel fully matures. That gives you a reason to start testing now, even if today’s revenue per session doesn’t yet justify a larger investment.
The measurement period changes the picture
A study’s publication date is often months after the behavior it describes took place. For example, research published in Marketing Science in April 2026 analyzed data from 973 ecommerce sites spanning roughly August 2024-July 2025. The main analysis found that ChatGPT referrals converted better than paid social but worse than the other established channels studied.
In several alternative analyses, the conversion difference between ChatGPT and organic search was no longer statistically significant.
According to Adobe's retail data, revenue per visit can change substantially over a year. In July 2025, non-AI retail visits were worth 128% more than AI-referred visits. By July 2026, AI referrals were generating 53% more revenue per visit than non-AI traffic.
The Marketing Science paper and the Adobe report cover different periods and samples, so each finding is unique to the dates studied.
Before comparing your results with a published benchmark, check who the study measured, what counted as a conversion, which channels were observed, and what dates were covered:
Check
What to look for
The audience
Businesses and buying journeys similar to yours
The definition of conversion
The specific action counted as success
The baseline
Organic search, another individual channel, or all non-AI traffic
The time period
The dates the visits happened, not the date the report was published
We don’t have an industry-wide referendum; AI referrals outperform other channels in some studies but fall behind in others.
For your business, you need to know how much AI visitors are worth to your site before investing more in AI visibility.
Why can AI referral volume understate the channel’s value?
AI referral counts can understate the channel's value for two reasons:
- The AI-referred visitors you’re able to count arrive further along in the buying journey compared to other channels
- Some AI-influenced visits are counted as direct or organic traffic
AI traffic is small but growing quickly
Semrush’s channel-mix study covered billions of visits across more than 50,000 websites. AI traffic accounted for just 0.14% of total visits but grew 66% during 2025.
The traffic share tells you the channel’s size, but to decide whether a small channel justifies more investment, you need to measure the business each visit generates.
AI visitors arrive further along in the buying journey
Shopify’s Q1 2026 data found that more than half of AI-referred sessions began on product detail pages, compared with about 20% for organic search. Shopify attributes the pattern to “journey compression,” which occurs when shoppers research their options inside an AI conversation before visiting a store.
Microsoft Clarity’s analysis examined 30 billion sessions across its 2,000 projects that received the most LLM traffic. Compared with organic search, AI referrals recorded 21.6% fewer quick backs, meaning fewer instances of visitors opening a page and immediately returning to the previous one. AI sessions also lasted longer while covering fewer pages.
If we judge the findings by page views alone, AI-referred visitors appear less engaged. Microsoft interprets the combined behavior as more focused interaction, however:
“AI-referred sessions follow a consistent pattern: Users spend more time within a session, but they do so across fewer pages,” says Annika Tanner, Product Manager at Microsoft AI.
The referral reports miss some AI-influenced visits
Carlos Silva, Content Production Expert at Semrush, notes that AI clicks without referrer information can appear as direct traffic. AI platforms don’t always pass along information that identifies where a visit came from. Without that information or some other tracking signal, GA4 can classify the visit as direct.
A buyer can also discover your brand through AI and later search for it by name. If they click an unpaid Google result, that visit can appear as organic search in GA4, excluding the earlier AI recommendation from the reporting.
GA4 classifies Google AI Overviews and AI Mode visits under Organic Search, excluding them from its AI Assistant channel. And in Semrush’s consumer survey on the modern shopping journey, 50% of surveyed US AI users reported purchasing after using AI for research.
The survey measured AI’s role in purchase research. Your referral report measures identifiable clicks. Some AI-influenced purchases will inevitably sit outside that report due to gaps in tracking capabilities.
How to measure AI traffic quality
There are four steps to measuring the quality of AI traffic:
- Identify AI referrals in your analytics
- Compare AI-referred conversion performance with other channels
- Review the pages AI-referred visitors land on
- Compare those results with where your brand appears in AI results
The measurement also includes comparing how often AI crawlers read your site against the visits they send back. We’ll start with the data you already collect and fill any tracking gaps along the way.
1. Identify AI referrals in your analytics
Celia Harding, founder of the LLM-visibility agency Leoprd, gives us a starting point: “Use GA4 to see if you're getting any traffic from AI models.”
GA4 now runs that first step for you. When it recognizes an AI assistant in a visit’s referrer data, it automatically sets the medium to “ai-assistant” and groups the session under AI Assistant in Default Channel Group reports. The campaign field also gets the label “ai-assistant.”
Open “Reports” > "Acquisition” > “Traffic acquisition” and select “Session default channel group” from the table’s dimension drop-down. Look for the AI Assistant row. You’ll find sessions and engagement data, plus key events and revenue if you’ve configured those measurements.
Convenient. Also incomplete.
Visits without referrer information can appear as Direct in GA4. Clicks from Google AI Overviews and AI Mode stay in Organic Search.
Next, check for AI referrals outside the AI Assistant channel. Switch the table’s dimension to “Session source/medium” and look for AI domains, likechatgpt.com or perplexity.ai. If an AI source appears within a medium such as referral, you have identifiable AI traffic sitting outside the default AI channel.
You can bring those visits together in your reporting with a custom channel group. The same approach lets you regroup historical AI referrals where source data exists.
Go to “Admin” > "Data display” > “Channel groups.”
Click “Create new channel group” and give the group a name. Then click “Add new channel” and name it “AI Referral Traffic.”
Set up two condition groups joined by “OR” so a visit qualifies when either condition matches:
- Default channel group: Select “exactly matches” and enter “AI Assistant”
The first condition includes traffic GA4 already recognizes as AI. The second adds visits from the domains you specify.
For example:
^(.*\.)?(chatgpt\.com|openai\.com|perplexity\.ai|claude\.ai|gemini\.google\.com|bard\.google\.com|copilot\.microsoft\.com|grok\.com|deepseek\.com)$
Check the expression against the source names recorded in your property, and update it when new AI referrers appear.
Save the channel, then move it above AI Assistant, Referral, and Organic Search. GA4 assigns each visit to the first matching channel, so the order is important.
Save the group. Return to the Traffic acquisition report and select your new group’s session dimension from the table’s drop-down. Your “AI Referral Traffic” row will show visits that match either condition.
View AI-referred sessions in Microsoft Clarity
Open Clarity’s dashboard and go to the “Channel” card. Select “AIPlatform,” which Microsoft describes as its AI platform channel. You can isolate AI-referred sessions and review recordings to see what visitors did.
A separate “Paid AIPlatform” channel will appear for visits from ads within AI experiences.
Filter for AI referrals in Shopify Analytics
Shopify merchants can check AI referrals inside Shopify Analytics. Go to “Analytics” > “Reports” > “Total sales by referrer.”
In the “Filters” panel on the right, select “Order referrer name” and set it to “is chatgpt,” as shown below. The resulting report displays orders and sales attributed to ChatGPT. Repeat the process for other AI platforms.
To check visits, use “Sessions by referrer” and filter by “Referrer name.” Keep the date range consistent across the traffic and sales reports so you can assess both over the same period.
Compare AI referral traffic with competitors in Semrush
Celia Harding also recommends looking beyond your own site: “Then uncover your share of models versus your competitors.”
Semrush’s AI Traffic dashboard estimates which AI platforms send visitors to competing sites and how your domain compares.
From the AI Traffic dashboard, enter your domain and competing sites, then click “Analyze.” Set the date and market filters to match the period and audience you want to compare across sites. The Traffic Distribution chart breaks down estimated visits by AI platform.
For example, ChatGPT is the biggest AI traffic source shown for H&M, ASOS, and Zara. H&M leads in estimated ChatGPT visits, followed by ASOS, with Zara trailing both. For Zara, the next move is to check which competitor pages appear in ChatGPT answers.
We’ll cover that visibility analysis in step four below.
The number of AI-referred visitors settles the size question. Next, let’s look at how many of your AI visits become sales or qualified leads.
2. Compare how AI visits convert against other channels
Per visit, how much business does AI traffic produce compared to your other channels? Use the same key event and reporting period for every channel so you can fairly compare results.
Pick the key event first. For an online store, this is likely a purchase. For a SaaS company, it could be a trial start or demo request. For a lead-based business, it might be a form submission, followed through to sales acceptance in your CRM.
Once you’ve identified the key event, open the Traffic acquisition report in GA4 with your custom channel group selected. Compare AI Referral Traffic with Organic Search, Paid Search, and Direct. Find the “Session key event rate” and select the event you chose. Check the session counts alongside the rates.
Read the conversion rate alongside revenue and order value:
Metric
How to calculate it
What it tells you
Session key event rate
Sessions with the selected key event ÷ total sessions x 100
How often a visit includes the action you’re measuring
Revenue per session
Attributed revenue ÷ sessions
How much revenue each visit generates
Average order value
Order revenue ÷ orders
How much buyers spend per order
For session key event rate, count each session once, even if the visitor completes the action several times. GA4 calculates the metric on that basis.
Revenue per session is the metric to watch. Siege Media found that AI traffic’s conversion advantage didn’t consistently translate into higher revenue per session. A channel with a higher conversion rate and lower revenue per session is sending buyers who convert on cheaper products or offers with thin margins.
For B2B, follow AI-referred leads into your CRM. Compare qualified-lead rate — qualified leads divided by total leads — and pipeline value per lead across channels. Use the same qualification criteria, and give each group the same time to progress.
Give the sample time
When AI referral volume is small, a few purchases can dramatically move the rate. SEO consultant Ana Precup shared results from one of her ecommerce projects in our AI visibility ROI guide: ChatGPT drove 471 sessions and 12 purchases worth roughly $2,000 in a month.
That conversion rate is 2.5% based on 12 purchases from 471 sessions. It'd only be 2.1% with 10 purchases. The two additional orders make a noticeable difference.
The session counts were far below Google organic, but as Precup said, “the commercial value of each visit was much stronger.”
So, compare over 90 days rather than just 30. Use the same dates for every channel, and check whether a promotion or one large order explains a jump.
Run the check in Shopify or Clarity
Shopify merchants can return to the traffic and sales reports from step one. Divide sales attributed to ChatGPT by ChatGPT sessions, then repeat for your other sales channels.
Use the same dates, and check how each report credits the source of a visit or sale. Both reports need to count ChatGPT referrals the same way.
In Clarity, filter for “AIPlatform sessions” and check your smart events. Turn on “Show percentage of sessions” to see how often AI visits include the action you’re tracking. Switch to another channel and compare the same event over the same dates.
Clarity adds what GA4 doesn't: Watch recordings of where AI sessions stall when they convert below expectations.
If AI visits convert well but revenue stays flat, check which pages bring in those visitors and what they buy, which also happens to be the next step.
3. Audit the pages AI-referred visitors land on
A channel-level number tells you whether AI-referred traffic converts. A page-level number shows you where it stops and starts.
Compare landing-page performance in GA4
To review AI conversion performance at the page level in GA4, open “Reports” >“Engagement” >“Landing page.” Click the “+” beside “Landing page” in the table to add a secondary dimension.
In the menu shown below:
- Select “Traffic source” in the left column
- Under “Session-scoped” in the middle column, select “Cross-channel”
- Select “Session AI Referral Traffic” at the top of the right column. If you named your custom group differently in step one, look for “Session [your group name].”
Once the secondary dimension is added, enter “AI Referral Traffic” in the search bar above the table and press “Enter.” The remaining rows show where your identifiable AI-referred sessions began.
Review sessions, key events, and revenue for each landing page. If the “Session key event rate” is missing, an editor or administrator can add it through report customization.
Select the same business outcome you used in step two. The results cover sessions that started on each page. A visitor can land on a guide, move to a product page, and buy. Sort by sessions first, then compare conversion rates and revenue per session among pages with enough visits to meaningfully assess.
Next, compare AI and organic visitors who land on the same page, over the same dates.
Use the results to choose what to investigate. The high and low baselines reflect comparable pages on your own site.
AI sessions
Conversion rate
What to do next
High
High
Keep the page current. If revenue per session is strong, test ways to attract more AI referrals. If it’s average or weak, check whether conversions represent valuable actions and whether visitors buy lower-priced or fewer products.
High
Low
Check the content for alignment with visitor intent, and watch recordings for obstacles to conversion
Low
High
Confirm the result holds beyond a few conversions, then investigate opportunities to increase visibility
Low
Low
Review the page’s purpose and sample size before investing in changes
Check which competitor pages receive AI traffic
Extend the review to your competitors. Open “Traffic & Market” >“Top Pages” in Semrush. Sort the table by “AI Traffic” to find pages with the most estimated AI traffic. Use the URL search to narrow the results to a section of the site.
In the Zara example below, there are 151 AI-referred visits to the Korean shopping-cart page, 100 to returns instructions, and 72 to a payment help page. The returns page receives more estimated AI traffic than organic search traffic: 100 visits versus 83.
The same report for H&M, in the screenshot below, shows that its UK sale page receives an estimated 92,200 AI visits, ahead of the UK homepage’s 60,400.
Zara’s filtered results feature shopping support, while H&M’s leading page takes visitors straight to discounted products.
For your own comparison, use the same reporting dates and market, then compare equivalent competitor pages. If a competitor’s sale or collection pages attract AI visits while yours receive few, carry those URLs into the citation check below.
Check which pages get cited by AI platforms
Now bring in Semrush’s AI Visibility Toolkit. Open “Visibility Overview,” enter your domain, and choose the AI platform and market.
Scroll to “Topics & Sources” and select “Cited Pages.” The resulting table lists cited URLs and the number of prompts in Semrush’s database associated with each page. Expand a URL to inspect the associated prompts.
Take Alo Yoga’s example below. ChatGPT answers cite 6,700 Alo Yoga pages. In the table, the homepage is cited for 962 US prompts, the help center for 50, and the hat collection for 47.
ChatGPT is drawing on both shopping and support content for Alo Yoga. The next check is determining whether citations to the hats collection bring in visitors who buy.
Look up the cited URLs in your GA4 landing-page report. A page with frequent citations but few identifiable AI visits needs a different investigation than a page receiving visitors who rarely convert.
For the first, inspect the associated prompts and answers citing the page. For the second, investigate what happens after visitors arrive on the page.
Check the information behind the recommendation
The audit also needs to cover the details AI platforms use to recommend you.
In Shopify’s Q2 research, shoppers referred by AIs using structured Shopify catalog data converted at twice the rate of shoppers referred by AIs relying on scraped or third-party product feeds. If you’re an ecommerce business, check the product information supplied to AI platforms.
Velvet Birdcage, a Shopify merchant, appears first in the ChatGPT response pictured below, replying to a prompt seeking a UK wedding dressmaker under $1,500. ChatGPT highlights their customization options and links to the website.
The store’s Valentine gown page, for example, is very specific in the product details. They say shoppers can add sleeves, shorten the dress, or change to a corset back. A measurement chart explains sizing, and ordering instructions tell shoppers to email their measurements, event date, and color choice.
For a B2B site, review the feature and integration details an AI answer could use to recommend your product.
4. Connect AI visibility to what happens after the click
Steps one through three measured the click and everything that followed. Step four explores the points where AI platforms decide if they're going to mention you, cite you, or send any visitors at all.
Check the scrape-to-referral ratio
Microsoft Clarity’s scrape-to-referral ratio compares AI crawling activity with the visits AI operators send back.
Open “Bot Analytics” in Clarity’s AI Visibility Dashboard and review the AI Scrape-to-Referral Ratio card. Check the domain coverage, then use the operator breakdown to see which AI sources return traffic.
You can open recordings filtered to a referral source from the card.
The scrape-to-referral ratio compares crawling with visits. A high scrape-to-referral ratio means AI bots are accessing your content often relative to the visits they send. If you see that pattern, check whether the pages getting crawled also appear in AI answers and attract clicks. The ratio flags a gap to investigate — conversion and revenue data tell you what the visitors who do arrive are worth.
And that’s where it stops.
