Monday review memo: four findings and what they are worth

All figures come from the HotelPulse warehouse. The bookings are the public Hotel Booking Demand dataset, replayed as a live PMS. The four problems below were injected into the simulation as ground truth, so this memo also shows that the platform finds them, and how close the dashboard estimates come to the truth.

Summary

# Finding Where it shows Estimated impact Action
1 A mobile payment outage (22 Sep – 5 Oct) pushed direct bookers to Booking.com Revenue › Direct share of new bookings; Website › Payment success rate by device, Failed payments by error code ≈ €3.5k extra commission in two weeks; ≈ €90k a year had it gone unnoticed Alert on payment success per device; escalate gateway_timeout to the payment provider
2 Since the cookie-banner redesign (17 Aug), web analytics sees only half of the website’s bookings Website › Tracking coverage by week Web KPIs understated by about 28 % compared with before the redesign Restore a balanced banner; report revenue from the PMS only
3 The resort’s new laundry supplier: +32 % towels laundered per room night since 11 Sep Operations › Usage per room night by item, Purchase suggestion ≈ €25k a year Check invoices against the contract; renegotiate per-towel billing or collection frequency
4 When the hotel is 90 %+ full, breakfast complaints are 2.8× as frequent Operations › Breakfast complaints by occupancy during the stay 189 nights a year are affected; extra breakfast staff on those nights costs ≈ €11k a year Staff breakfast from the occupancy forecast

Context for every euro below: OTAs bring 44 % of room nights at an average commission of 17.1 %, which is €948k a year. Direct channels bring 11.6 %.

1. Mobile payment outage: €3.5k in two weeks, €90k a year if unnoticed

What happened. From 22 September the website’s mobile payment step failed intermittently with gateway_timeout. Mobile payment success collapsed while desktop and tablet stayed normal. Guests did not give up on the stay: they booked it on Booking.com instead.

Evidence. Website bookings fell from 7.9 a day (the 8 weeks before) to 3.9 a day during the 14 days. That is about 56 missing direct bookings at an average value of €458, about €25.7k of bookings moved to OTAs. Booking.com intake rose from 51.9 to 59.0 a day over the same period.

Estimate. €25.7k × (17 % Booking.com − 3 % website cost) ≈ €3.6k extra commission. The simulation’s ground truth is 62 diverted bookings worth €23.7k, which gives €3.3k: the dashboard-based estimate is within 10 %. Repeated every two weeks, the same leak would cost ≈ €90k a year, about 10 % of the group’s total OTA commission.

Action. Alert on payment success rate per device type (a group-wide rate hides a mobile-only failure). Escalate gateway_timeout to the payment provider with the affected time window.

2. Tracking coverage: web analytics now sees half the bookings

What happened. The redesigned cookie banner made “Reject” the more prominent button. The tracker only sees visitors who accept.

Evidence. We match tracked bookings to the PMS by confirmation code. Coverage was 72 % before 17 August and 52 % since, which is below our 60 % reliability threshold.

Impact. Before vs. after comparisons of visits or tracked bookings are biased by about −28 % (52 / 72). A “drop in website traffic” since August is mostly a measurement effect. Rates such as conversion and payment success remain usable. No revenue is lost: revenue is always reported from the PMS.

Action. Give “Accept” and “Reject” equal prominence (consent stays free). Annotate dashboards with the change date. Keep reporting revenue from the PMS, never from the tracker.

3. Laundry supplier change at the resort: ≈ €25k a year

What happened. On 11 September the resort switched to a supplier that bills per towel and collects more often.

Evidence. Towels laundered per occupied room night went from 1.79 to 2.36 at the resort (+32 %), while the city hotel stayed at 1.79. Usage is measured per room night sold, so this is not explained by occupancy. The purchase suggestion shows the same change (+25 % against the previous 28 days) before the next order is placed.

Estimate. +0.57 towels × €0.85 = €0.48 per room night × 52,110 resort room nights a year ≈ €25k a year. About €3k has been spent so far.

Action. Check the invoices against the contract: billed towels per room vs. what housekeeping hands out. Renegotiate collection frequency, or reintroduce the towel-reuse card. Watch the Usage per room night chart for the resort after the change.

4. Breakfast at full house: staff from the forecast

What happened. Breakfast is staffed for a normal house. On nights when the hotel is (nearly) full, queues and empty buffets follow.

Evidence. We link each review to its stay through the confirmation code and compare it with the hotel’s occupancy during that stay. Breakfast complaints appear in 9.2 % of reviews from stays at 90 %+ occupancy, against 3.2–3.4 % at every lower level. A review that complains about breakfast scores 6.6 instead of 8.4. Both hotels together had 189 nights at 90 %+ in the last year.

Estimate. One extra breakfast attendant for 4 hours on those nights costs about 189 × 4 h × €15 ≈ €11k a year. That would avoid about 58 complaint reviews a year. The effect on the overall score is small (+0.03), but these are the nights with the most guests and the highest rates, so this is a guest-experience fix on the most visible nights.

Action. Use the booking pace (rooms on the books for the next 120 nights) to plan breakfast staff and buffet replenishment for nights forecast at 90 %+. Only 8 such nights are on the books right now (low season); the summer peak is when it matters.

Method and caveats