How we measure our data
Every figure on this site falls into one of five evidence types. Below are the benchmark sources behind the ROI model, and the plain-language labels we use so you can tell a measured result from an estimate or a target.
Benchmark sources behind the ROI model
OTA share of bookings
STR / Skift Research, 2025
50%
OTA commission rate
Booking.com / Expedia standard rate
15%
OTA-to-direct conversion
Skift Research: direct booking trends, 2025
2.1%
Guest vendor booking rate
Internal benchmark from pilot data
13.7%
Average vendor booking value
UK hospitality vendor network average
£45
Vendor commission rate
GSI vendor agreement terms
10%
Upsell revenue per room/year
Cornell Hospitality Q2 2025 report
£40
Guest review rate via Lucy
TripAdvisor industry benchmark, 2024
5%
Review-to-booking conversion
BrightLocal local consumer survey, 2024
1.5%
Lucy booking engine capture
Direct booking conversion benchmark, 2025
2%
Lucy booking engine fee
GSI direct booking engine pricing (vs 15% OTA)
5%
Staff hours redeployed/year
Internal pilot estimate (300-room baseline)
2,000 hrs
Loaded staff cost/hour
ONS UK hospitality earnings, 2025
£23
Software cost avoidance
Sum of 5 replaced SaaS subscriptions (see below)
£8,400
What our evidence labels mean
Industry benchmark
A figure taken from a named third-party source such as STR, Skift Research, Cornell Hospitality, ONS or TripAdvisor. The source is listed in the table above.
Internal estimate
A figure from our own pilot modelling or management projection. It is not a production measurement at a live hotel and is marked as an estimate wherever it appears.
Controlled test (smoke test)
A small, controlled technical test, such as a seven-category vendor booking test. It shows the mechanism works. It is not a full-scale load test and not a guarantee of live performance.
Target
A performance goal we are engineering towards, such as response times or uptime. It is not a measured result.
Code-complete, awaiting certification
The connector is built and sandbox-tested but has not been certified against the live system. This applies to Mews, Opera, Cloudbeds and Apaleo.
How we count features and project revenue
- Feature counts are taken from the platform's own registry of data types and backend functions.
- Guest-facing features (80+) are a subset of the wider operations platform (135 staff tools across 18 departments).
- Not every counted feature is live in production. Most are code-complete and sandbox-tested.
- Financial projections use the inputs shown in the ROI calculator so you can change any assumption and see the effect.