Why AI Traffic Shares Can Be Misleading ?

Eight times bigger. One tenth of one percent. Both are true, and you have only been shown one of them.

Shopify reported AI referral sessions growing more than 8x year over year. That number is on every deck about agentic commerce right now. What is almost never printed next to it is the share it multiplies.

One agency published both. Ten United States stores, 94 million sessions, 274 million dollars, two years of real analytics. ChatGPT, Perplexity, Gemini, Copilot and Claude, added together: 0.10 percent of sessions and 0.10 percent of revenue this year. Seventy-five thousand dollars out of seventy-six and a half million.

Now the part I respect them for. They also wrote down both ways their own number is wrong.

Too low, because a shopper who picks a product with a chatbot and buys three weeks later in a sale is recorded as direct or email. Their words: floors on AI's influence, not ceilings.

Too high, because all ten stores are their own retainer clients, already built for this. Their words again: a store that has invested nothing is likely seeing less.

Two arrows, opposite directions, on the same figure. So there is no honest point estimate, and anyone quoting you one is selling something.

Meanwhile, in the same dataset, the Shop app earned 35 times everything the chatbots referred. It is a switch in your admin and nobody is writing think pieces about it.

Four numbers from your own reports tells you where you actually sit. Then you can judge the pitch.

Follow Okiela for practical Shopify seller checks, tools, and ready-to-use templates.

#ai #shopify #Okiela