Three hundred boats, one marina, a six-week window where nothing can go wrong. That was the operational reality at Marseille during Paris 2024, and it's the kind of brief that exposes every weak point in how a fleet is actually run. Spreadsheets break. Radio nets get saturated. A single unreported fuel leak or a tender drifting off its mooring at 3 a.m. becomes a story everyone reads about the next morning. What the Olympic Marina taught us, working alongside the teams on the ground, is that fleet management at that scale stops being a logistics problem and becomes a software problem.
Why 300 boats changes the math
A harbour master running 20 boats can keep most of it in their head. They know which engine is overdue for a service, which skipper tends to leave the batteries on, which mooring has a worn line. Push that number to 300 and the human memory model collapses. You now have, on a typical day :
- Hundreds of engine hours accumulating in parallel, each on its own maintenance clock.
- Dozens of movements per hour during peak race windows.
- Mixed fleets : RIBs, support boats, committee boats, coach boats, each with different ECUs and sensor stacks.
- Multiple operators with overlapping responsibility but no shared source of truth.
The instinct is to throw more people at it. That works for a weekend regatta. It does not work for six weeks of competition where boats are running 12 to 16 hours a day. What you need is a layer that watches every hull continuously, surfaces only what matters, and gives every operator the same map.
What we actually deployed
Each boat got an Oria Box clipped onto its NMEA 2000 backbone (or NMEA 0183, where the older support boats hadn't been upgraded). No rewiring, no breaking the seal on factory installations. The Oria Box reads the engine ECU, GPS, battery state, and any other sensors already on the network, then pushes that over 4G to the Oria platform. The SIM is included, so there was no telecoms procurement headache, which on a project with this many stakeholders is not a small thing.
From the operator's side, the Oria app gave the marina team a live map of all 300 boats with :
- Position, heading, and speed in near real time.
- Engine hours per boat, with maintenance thresholds flagged before they became problems.
- Geofencing zones around the marina perimeter, the race areas, and exclusion zones.
- Battery voltage and charge state, which on RIBs that run electronics all day is the single most common cause of a no-start at the dock.
- Voyage replay : every track, every engine event, recoverable after the fact.
We've written separately about the deployment itself in our project debrief on Paris 2024, which goes into the install timeline. What's worth pulling out here are the patterns that emerged once the fleet was live.
Lesson one : maintenance becomes predictable, not reactive
The dominant failure mode in any working fleet is reactive maintenance. A boat comes back, somebody mentions it felt rough, the mechanic gets to it when they can, and meanwhile the engine hours keep stacking. At Olympic scale, you cannot afford that lag. A coach boat that drops out at 8 a.m. because nobody checked the oil hour counter is a sailor missing their warm-up.
With per-boat engine hours streaming in continuously, the maintenance plan shifts from "when somebody remembers" to "when the data crosses a threshold". You schedule oil changes the evening before they're due, not the week after. You catch alternator output sagging on a specific RIB before the battery dies on the water. You stop debating whether a service was done : the Oria platform shows the engine hour reading when the work was logged, and the next service date is calculated from there.
For owners running boats year-round, this same logic is what makes keeping a boat afloat through every season manageable rather than punishing. You stop sailing on hope and start sailing on data.
Lesson two : geofencing replaces half the radio net
Before fleet software, the question "where are our boats" was answered by VHF calls and binoculars. That gets old fast when you have 300 of them. Geofencing flips the question : instead of asking where each boat is, you ask the platform to tell you only when a boat does something unexpected.
Useful patterns we set up at the Olympic Marina :
- Marina perimeter : any boat leaving outside operating hours generates an alert. This is the anti-theft case, and it works whether the boat is being driven off or towed.
- Race area zones : safety boats entering or leaving their assigned sector were logged automatically, which made post-race debriefs much faster.
- Refuelling dock dwell time : if a boat sat at the fuel dock for more than X minutes, somebody knew. Small thing, but at peak hours it freed up the dock.
None of this replaces VHF discipline. It does mean the net stays clear for things that actually need a voice on the radio.
Lesson three : voyage replay ends arguments
Anywhere there are operators sharing equipment, there are disputes. Who hit the dock. Who ran the engine cold. Who took the boat into the exclusion zone during the medal race. With every track and every engine event recorded, those conversations get short.
This isn't about surveillance, and we said so at the time. It's about removing ambiguity. A skipper who knows the track is logged tends to drive more carefully, which is a safety win. A mechanic who can see the RPM curve leading up to a fault has a much better starting point than "it made a noise". And when an incident does happen, you have the actual sequence of events, not a recollection filtered through stress.
The same logic applies to ordinary recreational use. If you've ever wondered how often the most common incidents at sea involve a chain of small unnoticed events, voyage replay is what lets you reconstruct them and learn from them.
Lesson four : one platform beats five tools
The temptation, when you've got a complex operation, is to specialise. One app for tracking, one for maintenance logs, one spreadsheet for fuel, one WhatsApp group for coordination. Every Olympic-scale operator we've seen ends up there by default, and every one of them loses time stitching it back together.
The lesson from Marseille was that consolidating onto a single platform, with everyone seeing the same map and the same alerts, is worth more than any individual feature. The harbour master, the maintenance lead, the safety coordinator, and the team managers all looked at the same data, filtered for their role. Decisions got made faster because nobody was reconciling versions.
This is also why professional operators like Team Yachting have moved to a shared platform model for their own fleets. It scales down as well as it scales up : the same logic that handled 300 boats in Marseille handles a charter operation with 12.
What translates to a smaller fleet
You may be reading this thinking "I don't run 300 boats, I run one, or six". The honest answer is that everything above scales down cleanly. The Oria Box was designed for the recreational and small-professional market first. The Olympic deployment was a stress test, not a different product.
For a single owner, the relevant pieces are usually :
- Maintenance tracking tied to actual engine hours, not calendar guesses.
- Geofencing around your home port, with an alert if the boat moves when it shouldn't.
- Battery monitoring, especially if the boat sits unattended between weekends.
- Voyage history for your own records, insurance, or resale.
For a small charter operator, add fleet-level dashboards, customer-specific tracking, and a defensible maintenance log that holds up if a charterer disputes a damage claim. The mechanics are identical. Only the number of boxes on the map changes.
Paris 2024 mattered for us because it proved the architecture under load. Six weeks, hundreds of hulls, real consequences if anything went sideways, and the platform did the quiet work it was supposed to do. The question worth sitting with, whether your fleet is one boat or fifty : how much of what you currently track in your head, on paper, or in a group chat, would you rather have running on its own in the background ?
