CASE STUDY
Hundreds of hotels a year open on COMPASS.
Since 2021, teams at a global hospitality brand with over 9,000 properties have run a thousand hotel openings through COMPASS — each with its own critical path, its own budget, its own owner, its own handover date, and none of them waiting for the others.
- Partner
- A global hospitality brand with over 9,000 properties
- Scope
- Product design, engineering and operation
- Since
- 2021
- Product
- COMPASS
Where it started
1,000
It started with one question, and it was not about a platform. In 2021 a team asked whether the CapEx budget for a hotel opening could come from something other than a spreadsheet — the same spreadsheet, rebuilt by hand for every property, carrying rules that had never been written down anywhere else.
What we built first
A budget estimator. The arithmetic was never the hard part. A guest room in the Middle East has two telephones where a room in Europe has one, and behind that one difference sit different vendors, different prices and different lead times. The same is true of everything else a hotel is opened with: IT, FF&E, OS&E. Brand, country, owner type and room count each change what a property needs and what it costs, and they change it in combination, not one at a time. We moved those rules out of the spreadsheet and into the system, and the estimate started coming out of the project description itself.
How it grew
Once the catalogue existed, the next question answered itself. If rules can produce a budget from a project description, they can produce a task list from the same description — and a critical path from the task list. Then the tasks needed owners, and the owners needed a calendar, and the calendar needed to know who was already travelling or on leave. Projects were imported so their health could be read in one place; reading it made a forecast possible; a forecast is only worth the notes behind it, so project and call notes went in. Installations, training and the travel around them needed scheduling. Scheduling needed resources, resources needed absence data, and vendors, owners and senior management all wanted to see the parts that concerned them. By 2022 that was COMPASS V1, and it was running in departments across several regions.
2021
- Budget estimator
- Task catalogue
- Critical path
- Owners
- Calendar
- Absence
- Project import
- Project health
- Forecast
- Project & call notes
- Travel & training
- Resource scheduling
- External access
COMPASS V1 · 2022
One decision we didn't take
Every department wanted the same thing and meant something different by it. IT does not open a hotel the way Operations does, and neither works like Design or Finance. The obvious answer was to give each of them their own COMPASS — one per department, one per region, each shaped exactly to the team it served. We would have been building the same platform a dozen times over and maintaining it a dozen times after that, and the first person to ask how a whole region was doing would have got an answer assembled from twelve exports. So we did not do it. V2 was built the other way round: one platform, separated by department and region so that every team works the way it already works, and the separation deliberately breakable — because the person who needs to see every department at once is real, and they only exist if the walls come down on request.
Where it stands
V2 was designed and built through 2024 and 2025, alongside the V1 platform that kept running openings the whole time. It went into production in January 2026, and the global rollout is under way. A thousand openings have run through COMPASS across the two versions, in 146 countries, on more than twenty modules and a single data model.
What was hard
The hard part was never a single module. It was that the thing kept working. Every module we added made the next one obviously worth building, and a platform assembled that way reaches a size where the next addition costs more than it returns. We hit it at somewhere over twenty modules across several regions — the point where the honest options were to keep bolting things on, or to stop and rebuild the foundation underneath a system that had a thousand openings running on it. We rebuilt it. The second hard part was working out how much of what we had built was one department's way of working, and how much was everybody's.
THE BUILD
What it took
2.5 Mio.
Lines of Code
35,000
Hours Spent
>20
Modules and counting
50+
People Involved
5 Years
In Production
Everything above runs on one platform. Here's what's in it.