Why Some Visitors Book Direct and Others Leave

The direct-booking website is live. It is connected to the PMS, availability syncs, payments go through, and nobody has reported a broken page in months.

Direct share is still where it was two years ago.

That has become more expensive to ignore. In 2025 Airbnb moved hosts who connect through property management software from a split fee to a single host-only fee of around 15.5%. For most management companies, every booking that could have come direct and did not now costs noticeably more.

The usual response is to ask for a redesign, more traffic, or a new platform. Sometimes one of those is right. More often the problem sits underneath all three: the site was built for one generic visitor, and the people who actually book direct are several different visitors, each arriving with a different question.

This article is about finding those visitors, measuring them separately, and deciding what the site needs to do for each of them.

Start with a realistic ceiling.

Hostfully's 2025 trends data puts direct bookings at about 20% of reservations on average, and 25–40% for operators with 50 or more listings. PriceLabs describes established managers as typically landing around a third direct, with the rest spread across two or three marketplaces.

The reason is structural. Marketplaces are very good at reaching travellers who have never heard of you. Your website mostly is not, and spending to make it compete for cold discovery is rarely the best use of the budget.

What it indicates. Your direct bookings come mainly from people who already have a reason to choose you: guests who have stayed before, guests who found a property on a marketplace and then looked you up, referrals, and travellers searching for something specific enough that your portfolio is an obvious answer.

What to do with it. Stop judging the website on whether it out-competes the marketplaces for strangers. Judge it on whether it wins the guests who are already within reach. That is a smaller target, and a much more winnable one.

The marketing report shows one conversion rate for the whole site. It drifts down a little each year, and nobody can say why.

That number blends visitors who were never going to book: homeowners reading your management pages, job applicants, people reading a local guide, guests checking an existing reservation. It counts a week with no availability the same as a week with plenty. It misses the guest who browsed online and then rang to book. InterCoastal Net Designs makes a similar case in its piece on why year-on-year conversion comparisons are rarely like for like.

What it indicates. A falling sitewide rate can mean three quite different things. You may have a traffic problem: more of the wrong visitors. You may have a site problem: the right visitors cannot find what they need. Or you may have a measurement problem: bookings are happening, but not where the analytics can see them.

What to do with it. Before changing anything, split the number by the kind of visitor. The rest of this article is built on that split.

A management company's website usually serves five distinct visitors. They arrive by different routes, ask different questions, and leave for different reasons. Treating them as one audience is the most common structural mistake we see.

The map below is the useful part of this article. It is worth filling in for your own portfolio with your marketing and reservations teams in the room.

Two of these deserve a closer look.

The marketplace checker is the one most sites underserve. Hotels have long called this the billboard effect: a listing on a booking platform prompts some travellers to look up the business itself. When they arrive, they are not browsing. They are verifying. TouchStay's survey of 1,204 guests found that the deciding factor in booking direct was trust rather than discount, with around a quarter willing to book direct only if they trusted the operator. A guest checking you out needs evidence, which is covered in how to build trust before guests enter card details.

The homeowner prospect is the one that tells you who this map is for. A single-property host never has that visitor. A management company always does, because owner acquisition is how the portfolio grows.

Most management-company websites follow the same shape: a hero image, a search bar, a grid of properties, a property page, a booking engine.

That shape comes from the software, not from the guests. A website product sold to thousands of operators has to be organised around the one thing every operator has in common, which is inventory. So the structure becomes a catalogue.

For a small portfolio that is often the right call. With a handful of properties, a catalogue is easy to browse and a template is economical.

As the portfolio grows across property types and destinations, the catalogue starts to work against you. Every visitor gets the same entry point and the same grid, whatever they came for.

What it indicates. If your site underperforms despite good photography, clear prices and a working checkout, the limit is probably the structure, not the design. A new theme on the same catalogue will look better and behave the same.

"Know your guest" is familiar advice. It only becomes useful when it is specific enough to decide what pages the site should have. This is the sequence we use with clients, in the order we use it.

Start with your booking data

Export a year or two of reservations from the PMS. For each booking you want the booking source, property, length of stay, nightly rate, party size, lead time, and whether the guest has stayed before.

Group the bookings into segments that reflect how people actually travel: large family groups, couples on short breaks, guests with pets, month-long stays, returning guests. Then compare the segments on the things that matter commercially: rate, length of stay, repeat behaviour, and how much of your inventory genuinely suits them.

You are looking for two or three segments that are valuable, that your portfolio serves well, and that are under-represented in direct bookings compared with the marketplaces. Those are the guests the website should be built to win.

Then ask why those guests choose you

Data tells you who. It rarely tells you why. That knowledge sits with your reservations and guest-services team, who hear the questions and objections every day. A two-hour working session with them, segment by segment, usually surfaces more than any survey. The questions we start with:

• What do these guests ask before they book that the website does not answer?
• What nearly stops them booking?
• Which properties do they ask for by name, and why?
• What do they compare you with: other managers, hotels, the marketplace listing?
• What do they mention afterwards when the stay went well?
• Which facts decide the booking: bed configuration, parking, steps, distance, noise rules?
• When they book through a marketplace instead, what reason do they give?
• What could we offer direct that a marketplace cannot?

Turn each segment into pages, proof and a path to booking

For each priority segment, decide four things: the page type that serves them, the facts they need before a property page, the proof that answers their main objection, and the route to booking or enquiry.

A hypothetical example, for illustration only: a mountain portfolio finds that large groups book long stays at good rates, but almost entirely through marketplaces. The team says group organisers repeatedly ask about bed configuration, parking for several cars, and house rules.

The page-level details still matter: complete prices, honest property pages, clear policies. Our guide to what every property page should explain and the one on unexpected fees at checkout cover those. The structure decides whether the right guest reaches those pages in the first place.

One honest limit. If your portfolio has no real concentration, with no property type or guest need that a meaningful share of your homes serves well, segment-led structure will help less. In that case, start with the returning guest and the marketplace checker. Every portfolio has those.

This is a condensed version of the first step we run with clients. It will not give you a full picture, but it will tell you which kind of problem you have.

• Repeat guests. In the PMS, check what share of last year's returning guests rebooked direct, and how many came back through a marketplace.
• Branded search. In GA4 or Search Console, look at visitors who searched for your company or property names. These are mostly marketplace checkers and returning guests. Note how often they book.
• Landing pages. Group entry pages into home, property, collection or destination, owner, blog and careers. Note which groups bring visitors who book and which bring visitors who never do.
• Reconciliation. Compare direct bookings in the PMS with booking events in analytics for the same month. If the booking engine sits on another domain, check that cross-domain tracking works, as covered in why booking page changes make guests hesitate.

Reading the result. If returning guests drift back to marketplaces, the site has no route back for them. If branded visitors leave without booking, the marketplace checker is not getting a reason to stay. If non-branded visitors only ever land on the home page, the segment pages do not exist. If the PMS and analytics disagree by a wide margin, fix the measurement before fixing anything else.

The check usually points to one of three kinds of work, and they differ a great deal in cost.

Fix pages when the right visitors reach the right pages but stall on them: missing facts, surprise fees, weak trust signals. That is focused work on existing pages, and our series on why direct booking websites lose guests covers each fix.

Restructure when the visitors you want have nowhere to land: no collections for your priority segments, no returning-guest route, owner content tangled into guest navigation. This is strategy, information architecture and content, and it can often be done on your current booking engine.

Rebuild when the platform itself prevents the structure: it cannot hold collections, cannot carry property data cleanly from the PMS, or cannot be measured properly. If you go this way, protect the search visibility you already have, as described in how to redesign without losing search traffic.

Restructuring and rebuilding are where strategy, design and engineering have to work together, because the page structure, the booking engine and the PMS data all have to agree. That is the work we do on booking platforms, and the reason our longest short-term rental engagement, with The Right Property Group, is built around how the operator actually runs rather than a generic booking model.

A direct-booking site does not have one conversion rate. It has one for each kind of visitor.

Direct bookings come mostly from guests who already have a reason to choose you. A site organised as a catalogue answers none of them specifically. Decide which guests you want booking direct, learn what each one arrives needing, and build the structure around those answers.

If you would like a second pair of eyes, send us your site and the two or three guest types you want more of. We will show you where each one stalls and which pages are missing, as a short written or video review.