Is AI sending you customers? Three signals to check
By Daniela Ricaud, founder of Pretty CitableNo single report tells you whether AI is sending you customers. Look at three signals separately: AI referral visits in your analytics, what AI engines actually recommend when you test your buyers' questions, and what customers say when you ask where they first heard about you. Each one shows something the other two miss, so check all three before you move budget.
Signal 1: AI referral visits
When a shopper clicks a link inside ChatGPT, Perplexity or another AI tool, your analytics may record the visit as a referral from that tool. That's real, useful data. It also misses a lot.
Plenty of visits arrive with no clear referral source, and Google Analytics files those as "(direct) / (none)". Google lists missing campaign tags, redirects, typed-in URLs and ad blockers among the reasons. A shopper who read about you in an AI answer and then typed your name into Google later looks like any other visitor. So low AI referral numbers don't prove AI isn't influencing your sales, and you can't relabel your direct traffic as AI either. Nobody can tell which of those visits came from AI.
What it shows: clicks from AI tools that kept their referral data. What it misses: everyone who saw you in an answer and arrived some other way.
Signal 2: What AI actually recommends
Ask the engines your buyers' questions yourself and record the answers: which brands are named, whether yours is recommended or just mentioned, which sources are cited and where the links go. Here's the method.
Answers shift from day to day, so repeat your key questions on several days before you trust a pattern.
What it shows: whether you're on the shortlist when shoppers ask. What it misses: how many real shoppers asked those questions, and what they did next.
Signal 3: What customers tell you
Add one optional question after purchase: "Where did you first hear about us?" Let people answer in their own words. Wording and pre-written answer choices can steer how people respond, so a free-text box gives you a cleaner answer than a dropdown. If someone mentions an AI tool, follow up with two questions: which tool, and what they remember asking.
Keep each answer next to that order's analytics source. Don't overwrite one with the other.
What it shows: discovery in the customer's own words. What it misses: later steps. Someone who first found you on Instagram may still have asked ChatGPT to compare you with a competitor before buying.
Put the three together
- Referrals low, recommendations strong, customers mention AI: AI is likely helping more than your dashboard shows.
- Recommendations weak, competitors named: you have a visibility gap, whatever your referral numbers say.
- Recommendations strong, referrals land on the wrong pages: check where the links in answers send people, and whether those pages answer the buyer's question.
None of these signals alone proves AI caused a sale. Together, they tell you whether AI is a channel worth funding, and where the next dollar belongs.
Sources
- Google Analytics Help: understand (direct) / (none) traffic
- Ahrefs: How stable are AI citations? 37,500 citations tracked for 20 days
- Pew Research Center: writing survey questions
See where your brand stands
Run three buyer questions yourself in 60 seconds, or get the full AI Visibility Audit: 30 questions on four engines, a competitor map and a ranked fix list in 5 business days.