AI search, measured: common questions

No. SEO still matters, but showing up is no longer the finish line. When ChatGPT names a brand, the shopper usually gets there through a publisher's article and an affiliate link, not the brand's own page. In our Beauty study, only 46 of every 100 ChatGPT brand mentions led anywhere the shopper could buy that brand. The job now is turning a recommendation into a sale.

Neither one replaces SEO. GEO (getting your brand into AI answers) and AEO (getting your brand into the direct answer) sit on top of SEO, not in place of it. SEO gets your page found. GEO and AEO get your brand named by an AI. Neither one tells you whether a shopper could buy what was recommended. That is the last step, and it is the one Aqxle measures: from the AI answer, through the publisher and retailer, to the cart.

It can be, but only if you measure what happens after the mention. Most GEO reporting stops at visibility: how often an AI says your name. In our Beauty study, 54 of every 100 ChatGPT brand mentions were dead ends, with no shopping link to that brand on the page ChatGPT pointed to. Before you pay to be mentioned more often, check what happens after the mentions you already have.

Follow the whole route, not just the mention. That means tracing each AI answer to the article it quoted, every affiliate redirect, the retailer's page, and the seller who actually holds the Buy button. Normal analytics can't do this on its own. In our study, only about 3 in 100 links still showed an AI origin by the time they reached the retailer. Aqxle maps that route so the revenue can be traced back to AI.

For most brands, nobody can say yet, because the sales aren't being tracked. AI search ROI is the money from purchases that started with AI, minus what you spend to earn AI mentions, divided by that spend. The hard part is the money. In our Beauty study, only about 3 in 100 ChatGPT links still showed an AI origin at the retailer. So most AI-driven sales get logged as publisher, affiliate or direct traffic. Until the full route is measured, any AI search ROI number is a guess.

Use the standard ROI formula, (revenue − cost) ÷ cost, and build the revenue number from the route rather than from your analytics:

1. Mentions. How often an AI names you for real buying questions.
2. Buyable rate. How many of those mentions lead to a working shopping link for your brand. In our Beauty study this was 46%.
3. Retained sales. How many reach your own listing, not a reseller's. 39% of Amazon seller links in our study leaked.
4. Revenue. Retained clicks × conversion rate × average order value.

Then subtract what you spend to earn the mentions, including tools, content and agency fees. The fastest way to improve ROI is usually fixing the leaks after the mention, not buying more mentions.

Often not as directly as brands assume. In our Beauty study, 54 of every 100 ChatGPT brand mentions sent shoppers nowhere they could buy that brand. Among Amazon links that showed a seller, 39% went to a reseller or a listing nobody could buy from. A recommendation creates demand, but the route to the cart decides who gets the sale.

Usually the publisher, not the AI and not you. In our Beauty study, 7 in 10 ChatGPT shopping links credited the publisher through affiliate tracking. That means AI-driven demand shows up as publisher commission, and your analytics give the AI channel little or no credit.

Pick a fixed set of real buying questions and run them on a schedule. Record whether you're named, where you appear, and which sources the AI cites. Tools like Profound, Semrush and Similarweb do this well. Then add the step they skip: follow each citation to see whether it leads to a page where your product can actually be bought. Visibility tells you you're named. It doesn't tell you a sale is possible.

Track four things:

1. Mention rate. How often an AI names you for buying questions.
2. Citation sources. Which publishers and pages the AI leans on.
3. Buyable rate. The share of mentions that lead to a working shopping link for your brand. This was 46% in our Beauty study.
4. Credit and leakage. Who gets paid for the click, and how often the sale goes to a reseller or a listing nobody can buy from.

Most dashboards only cover the first two.

AI share of voice is your slice of all the brand mentions across a set of AI prompts. It's usually your mentions divided by total mentions, tracked by AI platform and over time. It's a useful starting number, but it counts mentions, not sales. Pair it with buyable rate, so a high share of voice isn't sending shoppers to dead links or resellers.

Because the AI tag usually disappears before the shopper arrives. In our Beauty study, nearly every link in a ChatGPT answer carried a ChatGPT tag (4,082 of 4,084). Only about 3% still showed an AI origin at the retailer, and none did at Sephora, Ulta or Nordstrom. Publishers and affiliate networks rewrite the link along the way, so the sale gets logged as publisher or affiliate traffic.

In GA4, create a custom channel group that matches AI referrers and sources such as chatgpt.com, perplexity.ai, gemini.google.com and copilot.microsoft.com. That catches visitors who click straight through to your site. It misses the bigger flow that goes through publishers and affiliate links. In our study, 75% of ChatGPT shopping links passed through an affiliate network. Measuring that traffic means following the links themselves, not just reading your analytics.

An AI visibility audit checks how often and where AI assistants mention your brand. Aqxle's route audit starts there and then follows each mention all the way to the cart. It checks the cited publisher, every shopping link and redirect, the retailer's landing page, and who holds the Buy button. You find out where the sale leaks: to another seller, a dead link or a reseller.