Skip to content
← Blog
Search and websites ·

Big brands lead AI search on price. Smaller sites can still be found

New BrightEdge research on travel shows big names dominate the booking and bargain questions in AI answers, and hold less ground on the advice and ideas questions that come earlier.

A small business desk with a laptop showing search results, notes and printed drafts, planning how to be visible in AI search

A small business website is unlikely to beat big brands on "cheap", "deal" or booking questions in AI answers. The realistic way to get visible in AI search is earlier in the decision. That means the ideas, advice and comparison questions people ask while they're still working out what they want. In new research, those were the questions where big brands held less ground. Answer them well, answer them often, and answer them accurately from what you actually know.

The research comes from BrightEdge, which looked at holiday travel questions in Google AI Overviews and ChatGPT. Search Engine Journal reported it on 31 October 2026. It covers travel, and only travel. The pattern behind it is still useful to anyone who runs a small business site, a member organisation's site or a set of client sites.

What the research found

In travel, the leading brands dominate AI answers on both engines. The leader differs by sector, though. American Airlines, Royal Caribbean and Expedia lead their categories on both Google AI Overviews and ChatGPT. Hotels split: Hilton leads on Google AI Overviews at 28.3%, while Marriott leads on ChatGPT at 29.4%.

The gaps between brands also change from one engine to the other. Expedia's visibility on Google AI Overviews is 2.2 times Kayak's (31.3% against 14.4%). On ChatGPT the gap narrows to 1.4 times. American Airlines leads airlines at 26.6% on Google AI Overviews and 25.3% on ChatGPT, ahead of United by 6.1 points and 9.1 points. Royal Caribbean leads cruise lines at 24.8% and 28.8%. Royal Caribbean and Carnival had higher visibility in the ChatGPT sample, and Viking had lower.

One detail matters for reading these numbers. As Search Engine Journal explains it, BrightEdge calls its figures share of voice, because AI answers don't have traditional ranking positions. A percentage here is how often a brand appears across the tracked answers. It isn't a place in a list. BrightEdge's category tables cover hotels, airlines, booking platforms and cruise lines, and leave out non-travel platforms such as YouTube and Reddit.

Where big brands are hard to beat

Price questions belong to the big platforms. Booking platforms appear in 78% of tracked Google AI Overviews answers involving terms such as "cheap", "deal", "budget" or "last-minute", against 17% of other answers. Expedia alone appears in 44% of the budget-related answers.

The same holds when someone is ready to book. On Google AI Overviews, 38% of booking-intent prompts name a booking platform. For inspiration prompts the figure is 12%.

BrightEdge adds its own caveat, and it's worth keeping. This is an observed association: using those words doesn't cause a brand to appear. So the finding tells you where the big names already sit. It doesn't hand anyone a formula for joining them.

For a small business, competing head-on for price and booking questions is the weakest use of its time. The brands that win there have scale you can't match with a few blog posts.

Where smaller sites have room

Booking questions are a small slice. Only 9% of the tracked Google AI Overviews prompts fell into the booking-intent category. Most questions people ask are something else.

On those other questions, the sources shift. YouTube, Reddit and AAA were more prominent in inspiration questions. Hotels and cruise lines featured more heavily in comparison questions. Booking platforms showed up more when the question was about booking.

The Search Engine Journal author reads this as an opening. In that reading, user-generated content and content drawn from human experience tend to do better on inspiration questions, with less big-brand dominance. That leaves room for sites like Reddit and YouTube, and possibly for travel blogs and smaller sites. Be clear about what that is: the author's interpretation. BrightEdge didn't say why big brands show less on these questions. And there's no data in the piece on how small sites actually perform in AI answers.

Even so, it's the sensible bet. The questions with the least big-brand dominance are also the ones a small business is best placed to answer from experience.

What this means outside travel

The research doesn't cover UK trades, professional services or member bodies. The shape of the questions carries across, though. In almost any sector, people ask two kinds of question. Some come early: what should I do, what are my options, which is better for me. Others come late: who's cheapest, who can come tomorrow, who's nearest.

Large directories, comparison sites and national chains are built for the late questions. The early ones are where your own knowledge counts.

Take a hypothetical tree surgeon. "Cheapest tree surgeon near me" is a price question, and the big listings sites are built to answer it. "Should I prune or remove a tree that's leaning towards the house?" is an advice question. A tree surgeon who has seen hundreds of leaning trees can answer it properly, in plain words, with the details a homeowner needs.

Now take a hypothetical trade body. Its members aren't asking AI where to buy cheaply. They're asking what a change in their sector means for them and which option suits a business like theirs. A trade body that explains those changes clearly, from the source's own facts, is answering comparison and advice questions on its own ground.

For an agency running several client sites, the planning question is the same for each client. Which early questions do their customers ask, and which of those can the client answer from real experience? That list is the content plan. The price questions can be left to the platforms.

What to publish, and how often

These steps follow directly from the pattern above.

  • Answer real advice and comparison questions. Write down the questions customers or members actually ask before they're ready to buy. Give each one its own piece, answered from what you know first-hand.
  • Keep the site current. If your news page hasn't moved in months, a steady run of short, accurate pieces does more than one long article a year. Pick a pace you can keep, such as a few pieces a month, and keep to it.
  • Write from a source's own facts. When a piece covers news, base it on the original source, link it and credit it. Don't add figures the source doesn't give.
  • Have a person approve every piece. Thin, generic AI text is the opposite of the human-experience content the author describes. A person who knows the business should read each piece, correct it and approve it before it goes live.
  • Don't stuff pages with "cheap" or "deal". BrightEdge's own caveat covers this. The link between those words and booking platforms is an association, so adding the words to your pages won't put you in those answers.

Accuracy matters more here than speed. An advice piece that gets a detail wrong does more harm to a small business than no piece at all. Check each claim against its source, and cut anything you can't support.

Mistakes to avoid when reading AI search research

Research like this is easy to over-read. A few cautions apply to this study and to the next one like it.

Share of voice isn't a ranking position. Hilton's 28.3% means it appeared in that share of tracked answers. It doesn't mean it sat at the top of a list.

An association isn't a recipe. BrightEdge says plainly that price words don't cause a brand to appear. Treat any similar finding the same way until someone shows otherwise.

Travel figures are travel figures. Don't assume the percentages transfer to your sector. Don't treat the opening for small sites as proven, either: it's one author's reading, and there's no data yet on how small sites perform.

Know what the study leaves out. The piece gives no sample size, no number of prompts, no data period and no methodology. It doesn't say which versions of Google AI Overviews or ChatGPT were tested. That doesn't make the findings wrong. It does mean they're a direction to test, not a rule to follow.

The sound approach is the patient one. Publish useful, accurate pieces on the questions you can answer best. Watch your own enquiries and traffic over several months, and judge by what you see. Keep a person approving everything that goes out: what appears under your name is your decision.

Written with AI by Corantr Media Desk from its sources' facts, checked claim by claim, and approved by a person before it was published. How we write.

More from the blog