A boutique hotel in Fitzroy charges $310/night for a King Loft. Last year it ran 71% occupancy across 22 rooms — call it 5,700 room-nights. Roughly 65% of those bookings came through Booking.com, Expedia or one of the smaller OTAs. Booking.com's cut on a hotel that size is 17%. Do the maths: that's around $197,000 in commission, every year, walking out the door — for the privilege of having other people sell rooms the hotel already owns.

The website was supposed to fix this. It was built in 2018 by a Melbourne agency that charged $14k for a Squarespace build, dropped a SiteMinder widget into a page, and called it a direct-booking channel. It converts at 0.8%. The OTAs convert at 4–6%. The maths of the leak gets worse every quarter.

The OTA tax

The OTA model is structurally brilliant for distribution and structurally toxic for margin:

Booking.com: 15% standard, 17–20% with Preferred Partner, up to 25% for Genius rate parity.
Expedia: 15–20% depending on contract tier.
Smaller OTAs (Agoda, Trip.com, Hotelbeds): 18–25%.

On a single $300/night room at 200 nights/year through OTAs, that's $9,000–$15,000 lost per room per year. For a 14-room boutique like the one we built The Luna around, the OTA tax exceeds $200k/year — a full-time employee, the marketing budget, a renovation, a salary.

OTAs aren't going away — discovery still matters. The strategic position is: let the OTAs do first-touch discovery, then convert the guest's next booking direct. Most hotel websites fail at the second half.

Here are the five fixes.

1. A real booking engine — not an enquiry form pretending

This is where 80% of hotel sites quietly lose. The "Book Now" button opens an enquiry form that emails the front desk. Reception replies in 24 hours with availability. By then the guest has booked on Booking.com — because the OTA let them confirm in 30 seconds at 11pm Sunday night.

That's not a direct booking channel. That's a contact form with a misleading label.

A real booking engine integrates with your PMS or channel manager and shows live availability with instant confirmation. The options that matter:

SiteMinder — the most common Australian channel manager. Solid booking widget, good rate parity controls.
Mews — cloud-native PMS with a clean API and modern booking flow. Increasingly the choice for new builds.
Cloudbeds — strong all-in-one for independent hotels, booking engine built in.
Little Hotelier — sister product to SiteMinder, aimed at smaller properties (under 20 rooms).
RMS — heavy in Australian holiday parks, motels, some boutique stock.
RoomRaccoon — newer European entrant, good UX, growing in AU.

The integration should do three things the enquiry-form version doesn't: show live availability confirmed in the same session, process payment at the moment of booking, and trigger the confirmation sequence automatically — instant email, calendar invite, pre-arrival reminders, upsell flow (spa, breakfast, late checkout).

If your "Book Direct" button opens a form, you don't have a direct booking channel. You have a contact form quietly losing every booking to Booking.com.

2. Room storytelling that justifies the rate

A $300/night room needs to earn its rate in the moment a guest lands on the page. Most hotel sites fail the same way: a grid of three identical interior photos, "Queen bed, ensuite, garden view", a price, a "Book" button.

That's a transaction. The guest mentally compares it to the same room on Booking.com, sees the same photos at the same price, and picks the channel they trust more — which is rarely yours.

The job is to make the direct version feel like a richer experience than the OTA listing. That's not more photos — it's story:

Large editorial photography — full-bleed, art-directed, shot at the right time of day. A bed in afternoon light hits differently than the same bed in flat noon light.
Slow scroll-reveal that lets each room have its moment — the bath in the corner of the loft, the morning light on the desk, the specific view from the specific window.
Specific detail copy — not "Queen bed, ensuite". "Carved iron bedstead salvaged from the 1898 building, blackout linen in indigo, desk facing the laneway so you can watch the morning coffee crowd."
Room-specific reviews — quotes that mention the actual room, not just "lovely stay".

A boutique hotel's rooms are not interchangeable. Treat them as individual products with individual stories — that's what guests at this price point are paying for. We treated The Luna as the demonstration: 14 rooms, 14 stories, full-bleed editorial photography. The contrast with a standard SiteMinder widget is the entire point.

3. Dining, bar and experience pages that could win a reservation alone

Guests don't book rooms — they book stays. The room is the bed they sleep in; the stay is the dinner, the cocktail, the morning at the spa, the breakfast on the terrace.

Most hotel sites bury this. The restaurant gets a mention on the "Hotel" page, a link to a PDF menu, a one-line description. The bar gets a photo. The spa gets a service list. None of it stands as its own destination.

The high-leverage move is to treat each non-room experience as its own page that could win a reservation independently:

Your restaurant page should be good enough that a Melburnian who isn't staying books a Wednesday-night reservation. Real food photography, current menu in HTML (not PDF), chef story, real booking integration (OpenTable, ResDiary, SevenRooms) that doesn't crash the brand.
Your bar page should pitch the room — cocktail list, bartender, live music night. Bar revenue from non-residents is pure margin most hotels leave on the table.
Your spa page should sell single treatments to walk-ins. A facials menu that ranks for "spa Fitzroy" is a meaningful revenue line.

Non-resident F&B revenue typically runs 40–60% of total for hotels that promote it properly, ~10–15% for hotels that bury it. The non-room pages are also top-of-funnel for the room business — a guest who comes for dinner becomes a guest who books a night.

4. A mobile-first booking flow that works on 4G cafe wifi

Most hotel bookings happen in transit. Phone, flaky 4G, between a coffee and a train, 90 seconds of attention before cafe wifi cuts out and the partner asks if it's booked yet. That's the actual context.

Most hotel sites are built for desktop and "made responsive" by squashing. The mobile booking flow then takes 6–8 seconds to load, the date picker doesn't render right, the payment form needs pinch-zoom, and the booking dies in step 4.

What mobile-first actually means:

Page weight under 1.5MB on room and booking pages. 2 seconds on 4G vs 8 is the difference between confirmed and switched back to Booking.com.
Tap targets at least 44×44px — date pickers, room selectors, "Book" buttons.
No autoplay video on mobile room pages — burns data, slows the load, achieves nothing.
Stripe / Apple Pay / Google Pay as default — three taps to confirm beats a 12-field form.
Proper input types and native date pickers, not custom JS calendars.
Tap-to-call in the header for the guest who wants to confirm something before committing.
No popups, no chat-bots, no "subscribe for 10% off" before the guest has seen the room.

The test: throttle to 4G in Chrome DevTools and try to book on your own site. More than five taps and 30 seconds means you're losing bookings — usually back to Booking.com. The wider pattern is in the real cost of a slow website — for hotels, every second compounds against your direct-booking margin.

5. Local-guide content so guests bookmark your site

Here's the move most hotels miss. A guest planning a Melbourne trip Googles "best restaurants Carlton" or "things to do near Fitzroy" — and lands on a Time Out listicle or someone's Medium post. None of which mention your hotel.

If your site has the best local guide to the neighbourhood — the actual places the actual concierge sends actual guests — the guest finds it via search, bookmarks the page, and books with you because the site already feels like the hotel.

What a real hotel local guide has:

Specific recommendations, not generic "things to do". "Walk five minutes to Lune Croissanterie before 8am to avoid the queue. The kouign-amann is the move."
Curated and opinionated — 15 places the hotel stands behind, refreshed quarterly. Not a list of 50.
Categorised — coffee, dinner, late-night, brunch, day trips, kid-friendly, rain plans.
Distance-from-hotel noted on each — "8-min walk", "12-min tram".
Owner or concierge voice — written like the GM recommending it to a friend, not a marketing intern.
Ranks for "[suburb] [topic]" searches — "best coffee Fitzroy", "Carlton dinner". Real SEO load-bearing content.

The Luna study includes a local guide as the demonstration — opinionated, hotel-voiced, distance-tagged. It's what separates a boutique hotel website from a directory listing.

What NOT to do — the hotel-site greatest hits

Things we see on Australian hotel sites every week that quietly drain direct bookings:

The "Book Now" button that opens an enquiry form. Discussed. Most common, most expensive mistake.
Stock photography of a white bathrobe and rose petals on a bed. Every hotel template ships this image. Guests register "this is a template" in half a second.
A "luxury" hero video on autoplay that takes 4MB to load on 4G and shows generic drone footage that isn't your hotel.
PDF menus from 2019 on the restaurant page. Same reason it kills restaurant sites — covered in Melbourne restaurant website essentials.
No room-specific pages — every room linked from a "Rooms" grid to a SiteMinder widget that loads in an iframe with its own typography.
"Award-winning" with no actual award named. Specifics or nothing.
A homepage hero that says "Experience luxury redefined." Means nothing. Specific beats generic.
No direct phone number on mobile. Some guests still want to call to confirm a detail. Tap-to-call should be in the header.
A "Best Rate Guarantee" hidden three clicks deep. This is the entire pitch for direct booking — show it everywhere. Booking.com's best rate should be matched (or beaten) on your site and the guest needs to know.

The pattern: a generic web agency built the site, treated it like a brochure, never thought about the OTA-vs-direct margin equation, and shipped a hotel website that looks fine and converts at 0.8%.

When you don't need us

Honest take: if you're a 4-room B&B with 80% repeat-guest occupancy and OTAs are 10% of your channel mix, the ROI on a custom site is modest. A Squarespace build with real photography and a proper booking widget will do the job for most of the year.

The moment a custom hotel site pays for itself is when OTA commission is a meaningful line on your P&L — typically 12+ rooms, 60%+ occupancy, 40%+ OTA share. At that volume, shifting 5–10% of bookings from OTA to direct earns back the build cost in 6–18 months and then keeps earning forever. We'll tell you honestly whether you're in that bracket.

The Pryce take — and a free audit

We build hotel sites differently because the constraints are different. Direct booking integration as the centre of gravity, not an afterthought. Room storytelling that justifies the rate. Restaurant, bar and spa pages that stand alone. Mobile-first build, fast on 4G, full booking flow under 30 seconds. Local-guide content that pulls top-of-funnel search traffic to your URL, not TripAdvisor's.

We wrote a dedicated page on hotel web design Melbourne that goes deeper on the build, the PMS and channel-manager integrations, and what a proper hotel site costs. It includes a free 10-point audit of your current site against every point above — OTA leakage, conversion leaks, mobile booking friction, written report, no pitch attached, keep the audit. If you want the wider build approach, see custom web design Melbourne.

To see what the finished product can look like, walk through The Luna boutique hotel design study — 14 rooms, dining, bar, spa, local guide, booking flow, built the way we'd build yours.

Related reading: The real cost of a slow website for small business — for hotels, "slow" maps directly to OTA commission you can't claw back. Every second of delay is a booking that defaulted to Booking.com.

How to Win Direct Bookings: 5 Hotel Site Essentials (2026)