Check these first
Six things that break at the seam
Each of these takes minutes to check on a phone and none of them requires access to an analytics account. Do them before commissioning anything.
- 01
The selection does not carry
Pick a room type and dates on the site, continue to the engine, and see whether both survived. Where they do not, every guest is asked to do the same work twice and a proportion decline.
- 02
The language resets
Do the same journey in Arabic. A booking engine that reverts to English, or keeps the language and loses the right-to-left layout, converts far below the English path and nobody has noticed because the rates are reported blended.
- 03
A second consent banner appears
Different domain, second banner, and for some guests a second decision at exactly the moment they were committing. Cheap to fix and invisible to anyone testing while already consented.
- 04
It is slow on a phone on mobile data
Test on a real device on a cellular connection, not on office wifi. Booking engines are frequently heavier than the site in front of them, and the guest is often on a phone.
- 05
It looks like a different company
A guest entering card details wants to be sure they are still dealing with the hotel. An unbranded engine at the payment step is a trust break, and trust breaks at the payment step are expensive.
- 06
A failed payment discards the reservation
Try a card that will decline. If the guest lands back on an empty search rather than on the payment step with the room held, the booking was lost to a retry that never happened.
Measuring it
Two domains means the funnel is broken in the reporting too
The booking engine usually sits on its own domain, which means the analytics on the hotel site stop where the engine starts unless somebody configured them to continue. Most properties therefore have two half-funnels and no view of the step between them — which is precisely the step that loses the most people.
Fixing the measurement is a small piece of configuration and it changes the conversation immediately, because a number replaces an argument. It is also worth doing before any engine replacement is considered: a property that cannot see the handoff cannot tell whether a new engine improved anything, and will be having the same discussion in two years with a different vendor.
What we do
Measure, configure, then judge the product
1 · Join the funnel
Cross-domain measurement so the handoff is visible, split by language and by device. Two to three weeks of data on most properties. A drop-off figure per step, per language.
2 · Fix the seam
Selection and language carried across, branding continuous, the second consent banner removed, the payment failure made recoverable. Configuration and integration rather than a project. The same funnel measured again against the same baseline.
3 · Then judge the engine
What remains after the seam is fixed is genuinely the product, and at that point a replacement decision has evidence behind it. Frequently there is much less remaining than expected. A replacement decision made on measurement rather than frustration.
Scope boundary
What we can change
What we do
- Join the funnel across both domains so the handoff stops being invisible.
- Work through configuration and integration before recommending a replacement.
- Test on real devices on cellular connections, in both languages.
The booking flow collects guest personal data and payment details across two domains, which makes the handoff a data-processing question as well as a conversion one. Confirm the retention and consent position for both parties before changing what is collected.
Primary sourceCommon questions
What buyers ask
Our vendor says the conversion rate is normal. Is it?+
Possibly, and the number they are quoting is probably an average across their whole customer base, which includes properties with very different demand sources. A more useful question to put to them is what proportion of sessions that reach the engine complete a booking, split by language and device — if they cannot produce that split, nobody involved can see the problem you are describing.
Can this be fixed without changing the website?+
Most of it, yes. The selection handoff, the language parameter, the branding of the engine and the payment failure behaviour are all configuration on one side or the other. What needs the site is the landing experience — a guest arriving on a room type page that does not show availability for their dates is a site problem, not a seam one.
Should we build our own booking engine?+
Almost never. The engine has to hold availability and rates in step with the property management system and the channel manager, handle payment, and keep working through every rate plan and promotion the commercial team invents. That is a product, not a project, and the cost of owning it exceeds what the seam fixes will return. Where a property genuinely needs something a vendor cannot do, build the layer in front and keep the engine.
In this cluster
A measured handoff, so the conversation about replacing the engine happens with evidence rather than with a hunch.
Join the funnel across both domains
