How people find us through ChatGPT

Shopify data · July–October 2026

1 in 5

online orders came from someone who found us through ChatGPT, September to 27 October 2026.

We’re a family florist in Albox, Almería, Spain, open since 1975. We deliver across mainland Spain.

Share of online orders that came from ChatGPT, by month, July–October 2026 Jul 2026: 4%4%Jul 2026 Aug 2026: 12%12%Aug 2026 Sep 2026: 21%21%Sep 2026 Oct 2026: 23%23%Oct 2026
Share of online orders from ChatGPT, by month. October 2026 runs to the 27th.
  1. 22%of online revenue, September to 27 October, came from ChatGPT orders
  2. 100%of orders from ChatGPT were from first-time customers
  3. 84%of customers arriving from ChatGPT landed on a city page
ChatGPT’s share of online orders and revenue. In-store sales and cancelled orders are excluded. Source: Shopify.
Month% of orders% of revenue
July 20264%2%
August 202612%11%
September 202621%18%
October 2026 (to the 27th)23%26%
Sep–Oct 202622%22%
Jul–Oct 202616%15%

How we counted

We reviewed orders placed on persaflores.com from 1 July to 27 October 2026. For each order, we checked the visit Shopify records before the purchase. If that visit came from ChatGPT, we counted it as a ChatGPT order. Shopify has no source recorded for some orders, so the actual share may be a little higher.

Who finds us this way

Just over half of the orders from ChatGPT were paid for from outside Spain. Just over half were placed in Spanish; the rest were in English. Here are three real orders, with no customer names:

  • Ordered from Ireland in English: a bouquet with a vase and a candle, delivered to a village in Segovia province, Spain.
  • Ordered from Bogotá in Spanish: sunflowers with chocolates, delivered to a village in Valladolid province, Spain.
  • Ordered from Germany in Spanish: roses and lilies with chocolates, delivered to Albox, the town where our shop is.

What to check in your online shop

Start a fresh ChatGPT conversation and ask the questions a customer would ask. Then check these things on your website.

  1. Keep the information consistent across your site. Opening hours, delivery times, prices and promises need to match.Why it matters: the assistant reads your whole site, not just the home page. If it finds two versions of a fact, it can’t tell which is right, and recommending someone else is easier.
  2. Check older pages too. Make sure the information is still accurate.Why it matters: to an assistant, an article from years ago carries the same weight as one from yesterday. An outdated detail can be the first thing it finds about you.
  3. Include your address, phone number and delivery area in every page’s structured data, in every language.Why it matters: it’s the one format a machine reads without interpreting. It learns where you are and where you deliver without guessing from the text.
  4. Let ChatGPT’s crawler read your delivery and returns terms. Check that those pages aren’t blocked.Why it matters: before recommending a shop, the assistant looks for what delivery costs, when it arrives and what happens if it fails. If it can’t read that, it recommends a shop that lets it.
  5. Show the full price, including delivery, where the customer decides to buy. Make clear who pays, who receives the order and what happens if something goes wrong.Why it matters: assistants compare shops on the delivered price. A price without delivery looks cheap until it’s compared, and then it drops off the list.
  6. Keep the options few and clear, with a price for each.Why it matters: with twenty sizes, the assistant can’t tell which price to quote and moves on to the next shop. With a few, it can give a price and recommend you.