Articles — Why you were skipped
Why you were skipped
You can be recommended and still not get chosen
The chatbot recommended your watch or your hotel, but it still had to guess whether someone could buy or book it, because the price, the availability, or the next step was missing.
Illustrative example, not a logged test.
You asked for a GPS watch under €300, and ChatGPT named Northloop’s Trail Watch.
You asked where to stay near the station, and it named a small hotel you had never heard of.
Both sounded like wins until you tried to act on them. The watch was “from €249” on a page that actually charged €279, and nobody said whether it shipped to your country. The hotel looked bookable in the chat, but the link went to a third-party listing with different cancellation rules than the operator’s site.
You can also land here after being skipped or named with wrong facts: you made the shortlist, but the answer was not safe to trust for a real decision. The overall pattern is in how AI chatbots build a shortlist.
In a log of the same question, recommended and described correctly are separate lines. This article is the line where you were recommended and the price, the address, or the way to buy or book was still not safe to act on. The rest of that log is in Those AI visibility tools just count mentions.
The one-page figure: Can someone actually pick this? — shop and travel side by side.
Recommended is not the same as ready to choose
Recommended, still not ready
Northloop Trail Watch — “from €249,” last week’s promo still in a feed row, no ship-to country, “available at Shop X” when Shop X never sold this SKU, the shop’s own product code, in your market.
Ready to choose
Northloop Trail Watch S, black, NL-TRAIL-S-BLK — €279 EUR, in stock, ships to your country, buy on the shop’s product URL with returns stated on the page.
Buying and booking depend on facts the chatbot can check. If those facts are missing, it guesses.
Four questions the assistant still has to answer
Trail Watch S in black, this product code, sold by Northloop. Use the exact product title that matches the feed, not a vague category name or a bundle label.
Ships to your country and the one next to it, with 30-day returns. Call out when a different case size is a different product code from the Northloop Trail Watch in the ask.
€279 on the buy button, in stock tonight. Match the live buy price; drop stale strikethrough “was” prices and old feed rows.
Buy on this product URL from this merchant. Point to a real checkout path instead of an invented “available at Shop X” from another region’s marketplace.
Shops and travel
For shops, name, price, stock, and identifiers are already the spine in your product lives in three places and wrong price, wrong shop, fake discount. A buying question still needs ship-to, returns, variants, and a shop the chatbot can point to without inventing one — the one-page figure lays out offer, limits, now, and next for both lanes.
For travel and local stays, stale hours and closed venues are covered in AI place answers come from different directories. The same frame adds who you pay, whether a booking site or the operator runs the reservation, and whether the dates in the answer match what is actually open.
Limits on the offer are what you publish. Constraint fit is what the asker typed in the question. You might not ship to their country. You might be closed on Mondays. The rate might be non-refundable after today. The next step is a buy or book URL on your site that you control, separate from checkout inside ChatGPT.
If you are the person asking
Before you buy or book from a named option, check ship-to or service area, returns or cancellation, tonight’s price or hours, and who actually takes your money. Treat a confident name as a hint. Verify the facts before you trust a sure-sounding answer, which is the same habit as in Why AI won’t just say it doesn’t know.
If you run the shop, place, or trip
Pick one hero product or one bookable offer. Write offer, limits, now, and next once in plain language, then make the webpage, structured data or feed, and directory listings agree. Fix the fact you can verify on a live page this week before you pay for a homepage rewrite.
Read next if: Find out why you were skipped before you rewrite the page · Those AI visibility tools just count mentions