This afternoon I crawled a client’s website. Every page, the way a search engine or an AI model would see it.
The client is a wilderness tour operator. Good people, real trips, years of happy guests. The site looks great to a human being. Beautiful photos, clear tours, a booking path that works.
Then the crawl came back. The site had 73 pages. About 40 of them were leftovers from the template it was built on: sign-in screens, sample pricing tables, five practice landing pages, a change log. All of them live. All of them listed in the map the site hands to search engines. Thirty-nine of them carried the same title: the name of the template.
To a machine, more than half of this tour company says it is a software company.
Yours might be saying something just as wrong. You would have no way of knowing, because you are not the one reading it.
I sat there for a minute. Nobody did anything wrong. The designer moved on, the owners never had a reason to look, and the site worked fine for every human who visited. The problem lives in the part of the website no person ever sees.
That part used to matter only a little. It is starting to matter a lot, and it matters to you.
When someone asks ChatGPT, Perplexity, or Google’s AI which outfitter to book, which firm to call, or what a service should cost, the machine answers the question itself. The buyer reads a summary first. Sometimes they never click through at all.
That changes the job your website does. It is no longer just the storefront where a buyer browses. It is the source the machine quotes from when the buyer never shows up at all.
Think of it like the post office. You can wrap a package beautifully, but nobody at the post office admires the wrapping or opens the box. The sorting machines read the label: who it is going to, where, who sent it, and whether the postage is paid. Get the label right and the package lands on the right doorstep. Leave it smudged or blank, and it sits in a bin somewhere or shows up at the wrong address.
Your website works the same way. The design, the photos, and the copy are the wrapping, built for people. Structured data is the label: plain code that states who you are, where you are, what you sell, for how much, and how long it takes. The machine skips the wrapping. It reads the label to decide what you are and who to deliver you to.
I call this the Shipping Label Problem. Most owner-operators have spent real money on the wrapping and never looked at the label.
This client’s label was blank. Not a single page carried structured data. No business name and address stated for machines, no tour prices, no durations, no seasons. So the machines guessed, and they guessed wrong. Google currently lists them at the hotel their office happens to sit inside. The package went to the wrong address.
Here is my prediction, and I am putting a date on it so you can hold me to it. By the end of 2027, more of your first impressions will happen inside an AI answer than on your homepage.
The businesses those answers name will not be the biggest ones. They will be the ones the machine can describe with confidence.
That is the good news. In the old search game, the biggest budget usually won. In this one, a small operator with clean, consistent facts can be named ahead of a large firm with a vague site. I wrote about how buyers default to the safest-looking name in The room we had no business winning. The machine has the same instinct. When it is unsure, it leaves you out.
AI search does not reward the loudest business. It rewards the clearest one.
So here is what we are doing for the tour operator, starting next week.
Cleanup comes first. We take the template leftovers out of the map and out of the index, and we give every real page its own title and description. No point writing a clean label on a box that is still covered in old labels.
Then we write the label. The business name, address, phone, and hours go on every page in plain code. All eighteen tours get marked up with price, duration, season, and what is included. The FAQ gets marked up so its answers can be quoted directly.
Then we ask the machines again and see what they say.
And here is my prescription for you. It takes about fifteen minutes.
Ask ChatGPT or Perplexity to describe your business, who you serve, and what it costs to work with you, so you see what the machine believes before your buyer does. Then type your web address followed by /sitemap.xml into your browser, so you see every page you are telling machines about. If anything in either one surprises you, ask whoever built your site where your structured data is, so you know whether anyone is minding the label.
If they cannot answer, you have found the Shipping Label Problem. The wrapping can wait. Fix the label first.
Or reply with your web address, and I will tell you what your label says.
Richard D. Stuart
The Weekly Prescription