Search-ready and AI-ready come standard on a HomeRunner site
More than a third of Americans planning a summer trip this year used an AI assistant to help plan it. That’s 37 percent in the Allianz Partners survey Ipsos ran in March and April, across 2,001 US adults. Adobe’s traffic data points the same way: visits to US travel sites that started in an AI assistant were up 194 percent year over year in May, and those visitors now convert almost as well as visitors from a regular search.
So the question for a vacation rental brand has changed a little. It used to be whether Google could read your site. Now it’s also whether ChatGPT, Claude, Perplexity and Gemini can. Most listing pages on the web are drawn in the browser by JavaScript after the page loads. AI crawlers don’t run JavaScript, so what they get is an empty shell, and what they can’t read they can’t recommend.
The search half is the part we’ve done for years, and it still comes first. Every property page is rendered on the server with its own title, description and clean URL, and it carries VacationRental structured data. The homepage carries the LodgingBusiness record Google expects from a lodging company. The site ships an XML sitemap that includes every property and the reviews page, and new or changed pages are pushed to Bing and the other IndexNow engines the moment they publish. At launch we carry the old site’s URLs over as redirects so nothing you’d earned goes dark, and the checkout and confirmation pages stay out of the index. Photos come through our image proxy resized and cached, so a listing with forty photos still loads fast on a phone. None of that is remarkable in 2026, which is the point. It’s the floor, and the AI pieces stand on it.
On top of that, here is what a HomeRunner site publishes for the assistants, included in the subscription, with nothing extra to buy.
- Property pages rendered on the server, with the schema.org VacationRental structured data Google documents for lodging: the home, its rooms and beds, the location, the offer.
- A machine-readable copy of every listing. One page per property with rates, availability and minimum-stay rules for the window you choose, an AI sitemap that lists them, and a site catalog that points an assistant at all of it.
- A robots file that welcomes the AI crawlers by name and tells them which use is allowed. If you’d rather stay bookable by assistants while saying no to model training, that’s one setting.
- A free, read-only MCP server for your site, switched on from the dashboard. Paste its link into Claude or ChatGPT and the assistant can check open dates against the calendar HomeRunner already syncs from your PMS, and come back with a dated total, or an indicative one where live pricing isn’t available.
- Receipts. The Insights Hub in your dashboard shows which AI bots read which listings, how many times, and which bookings started from an AI referral.
For the week ending October 22, the Insights Hub across our customer sites showed this.
- About 40,000 reads of machine-readable listing pages by AI crawlers in seven days.
- About 5,700 of those came from the search and agent bots, OAI-SearchBot, Claude-SearchBot, PerplexityBot, ChatGPT-User and Claude-User, which are the ones answering a traveler’s question in the moment. The rest were training crawlers.
- 94 percent of sites had at least one search or agent read in the week.
- Assistants asked for a live, dated price about 3,900 times.
Reads aren’t bookings. Booking Holdings has said AI-sourced bookings are still under 1 percent of its own. On our side, bookings that landed with a ChatGPT tag were rare through July, then more than quadrupled in August and held that pace through the first three weeks of September. On two of our larger customer sites, visits that began in ChatGPT already account for 2 to 3 percent of direct booking value, and nothing special was switched on to get there.
Which is why we don’t sell any of this as an add-on. We ran our own test in four markets this month, asking the four assistants where to stay and counting who got named. ChatGPT and Gemini named no independent vacation rental operator at all. Claude and Perplexity did, and the operators they named had more local content on their sites, local guides and pages written for each kind of guest. Structured data separated nobody. Neither the prospects nor the operators that got named carried it. Being readable gets you a seat at the table, and the content on top of it decides whether you get named.
That’s the case for treating it as table stakes. It’s cheap for us to do well once for everyone, and it’s the wrong thing to meter. So it’s on by default, included in the subscription, with the numbers in your dashboard.
The effort now goes into what sits on top. Local guide pages for the places guests ask about, each one on its own crawlable URL. A reviews page that puts every review on one page an assistant can read end to end. And on your property pages, plain answers to the questions guests actually type, is there a crib, how far is the beach, can I charge the car, written where an assistant can quote them. Readable is the part HomeRunner takes care of. Getting named is the work we’re on now.
How the pieces fit, in detail: https://homerunner.io/docs/ai-ready-how-the-pieces-fit/. The AI-Ready overview: https://homerunner.io/ai-ready/