Why Recommendations Matter More Than Rankings

For twenty years the question leadership asked of marketing was “are we visible?” — do we rank, do we appear, does the traveller find us. That question is not wrong, but it is no longer sufficient. Almost three in ten Australian adults now use an AI assistant to plan travel, and the leading uses are recommendations: where to go and where to stay.¹ A recommendation is not a list of ten results. It is two or three names and a reason for each.

Rankings and recommendations are different mechanisms

A search engine indexes the web and orders results by relevance and authority. The traveller then chooses. An assistant retrieves a handful of pages, decides what is true, and chooses on the traveller’s behalf. The traveller receives the conclusion.

Three things follow for an organisation.

Being present is not being chosen. In a ranking, position five still gets clicks. In a recommendation, there is no position five. You are named with a reason, mentioned as an afterthought, or absent — and absence produces no signal at all. No traffic falls; the traveller simply never knew you were an option.

The reason attached to your name comes from someone else. Assistants describe organisations from the sources they retrieve, and the published citation data shows those sources are overwhelmingly third-party: review platforms, booking intermediaries, regional round-ups.² The reason you are recommended — or not — was written by parties whose interest is the transaction, not the guest.

The link decides who gets the business. An assistant can recommend you and send the traveller to a reseller in the same sentence. In Sitter’s hotel study, roughly a quarter of one assistant’s links went to intermediaries rather than the hotel.² Search visibility never behaved like this; the click went to whoever ranked.

Why leadership should own it

It is tempting to file this under marketing. Three reasons not to.

First, it is a distribution question. Where the traveller is sent, and on whose terms, determines margin and the customer relationship. That is commercial strategy, not campaign management.

Second, it is a brand and risk question. Peer-reviewed research shows that a single inaccurate AI answer measurably lowers a traveller’s intent and trust, and that the traveller attributes the error to the organisation, not the assistant.³ An assistant misquoting your price or a withdrawn inclusion is making a promise in your name.

Third, it crosses every function. Marketing owns the website, sales owns the intermediaries, operations owns the listings, technology owns the feeds. The assistant assembles all of them. Nobody owns the assembled result unless leadership says so.

Four questions for the next leadership meeting

1. When a traveller asks an assistant for an organisation like ours, are we on the short list — and against whom?
2. Of the answers that name us, how many send the traveller to us rather than to an intermediary?
3. Which facts about us are currently wrong in public, and who is quoting them?
4. Which role is accountable for how we are represented across assistants, and what do they measure?

The good news is that none of this requires a new platform. It requires the discipline already applied to pricing and brand — know the facts, own the sources, check them on a cycle — extended to the sources assistants rely on. The first step is a measured baseline of what assistants say today.