When a shared operator, a corporate commuter pool or a municipal programme signs an e-bike order, the hardest-negotiated number is usually the unit price, and it is the number that predicts the least about what the fleet will cost to run. Availability, mean time to repair and the cost of keeping each bike earning per kilometre decide whether a deployment survives its second year. Procurement documents that score on purchase price alone tend to produce a fleet that looks efficient in month one and expensive from month twelve onward. Moving the scoring model toward cost per available kilometre is the highest-leverage change a buyer can make, and it is also what lets a dealer justify a better-specified bike instead of discounting a weaker one.

A fleet total-cost-of-ownership model has six lines, and only one of them appears on the order confirmation. The first is the vehicle itself, including landed assembly and pre-delivery inspection at the depot rather than at the factory gate. The second is the spare parts reserve: the buffer of batteries, controllers, wheelsets, locks and tyres held locally so that a repair is not a shipping event. The third is maintenance labour, best measured in minutes per bike per month rather than in workshop hours. The fourth is downtime, which converts directly into lost trips, missed deliveries or empty docks. The fifth is mid-life battery replacement, which arrives as a single large capital event roughly a third of the way through the asset life. The sixth is residual recovery, including whether used units can be resold, redeployed to lower-demand duty or parted out. For high-utilisation fleets, lines two through five usually outweigh the purchase discount being negotiated.

Availability only becomes an argument when it is converted into output. A 500-unit fleet running at 99% uptime has roughly five bikes unavailable at any moment; at 95% it has twenty-five, which is the difference between a dispatch plan that absorbs a failure and one that reshuffles routes every morning. The same arithmetic reads differently to different buyers. Shared operators price downtime in lost ride revenue and rebalancing labour. Corporate commuter fleets price it in employees who return to the car park after two failed attempts. Municipal programmes price it in service-level reporting and audit trails, where a bike off the road without a documented repair record is a compliance problem rather than only a cost problem. A single TCO template cannot serve all three, because the weighting of the downtime line changes with the buyer.

Hardware specifications that read as performance choices in a catalogue are, in fleet terms, serviceability choices. Internally geared hubs remove the derailleur hanger, cable stretch and indexing adjustments that generate most routine workshop tickets, at the cost of a heavier wheel and a more involved repair when something inside the hub does fail. Belt drives trade a higher initial cost for the absence of chain lubrication, stretch and corrosion, which matters most in coastal or winter-salted cities. On braking, hydraulic discs give consistent lever feel and pad wear that is visible without disassembly, while mechanical discs are simpler to adjust at the roadside with fewer fluids to manage. None of these is universally correct; the question is which failure modes your technicians can clear on site and which ones require a bench.

Frame-level design decides how long a repair takes. Single-sided axle mounts, quick-release wheels, tyres with puncture protection, integrated harnesses that disconnect at defined points, and controllers, battery locks and GPS modules that swap as complete units all shorten the job. The practical benchmark is a fifteen-minute intervention: the bike comes onto the stand, the failed module is identified, replaced and logged, and the bike returns to service before the technician moves on. Diagnostics support that pace. A CAN or serial bus that exposes fault codes, a controller that accepts over-the-air updates, and standardised connectors for lock and GPS hardware let an operator read a fault remotely before dispatching anyone. Standardised interfaces also protect the operator from being locked into one supplier for telematics or replacement electronics.

Environmental tolerance is usually decided by where the fleet is deployed rather than by price. Ingress protection ratings, salt-spray resistance and tolerance of high-pressure washing determine whether a bike survives a humid coastal city or a municipal wash bay. On the energy side, the choice between battery-swap cabinets and depot charging shapes the whole operating model. Swap cabinets reduce charging time to a battery exchange but require cabinet space, grid capacity and a battery stock that is always partly idle. Depot charging uses the batteries the fleet already owns but concentrates demand into a night window, needs electrical work at the site, and adds labour if staff must move bikes to chargers and back before the morning shift. A swap fleet needs fewer bikes per route but more batteries per bike; a charging fleet needs the reverse.

Battery management data is the cheapest maintenance tool in the fleet. State of charge and state of health reporting, cell-group voltage-delta warnings and remote power derating let an operator pull a weakening pack during a scheduled service instead of recovering it from the street. Retirement should be driven by accumulated cycles and observed capacity rather than by calendar age, because a bike working three shifts a day ages its battery far faster than one on a commuter schedule. Packs retired from primary service can move into lower-demand roles such as depot or staff bikes, and that second-life decision belongs in the TCO model from the start rather than being discovered at the end of year two.

Concentrated charging introduces requirements that do not move in a negotiation. Battery rooms and swap cabinets are subject to local certification, fire separation and insurance conditions, and those constraints can dictate site layout and how many packs may be stored in one place. The commercial side is equally concrete. Service-level agreements should state arrival times for critical components such as batteries, controllers, wheelsets and locks, the ratio of spare parts held per hundred bikes, and what the first-year replacement scope actually covers. Warranty terms need to separate field repair from factory return, define how misuse, vandalism and theft are treated, and say who supplies a substitute bike while a unit is off the road.

A pilot of fifty to one hundred units over ninety days generates the data that makes a large order defensible. Record failure rate by component, mean time to repair, maintenance minutes per bike per month and measured battery degradation, then use those figures to set the specification and the spare-parts package for the full rollout. Two clauses are worth negotiating before the pilot starts: an interface commitment covering GPS and lock APIs, plus ownership of ride and fault data; and an exit clause guaranteeing data export and continued spare-parts supply if the relationship ends. The procurement that ages well is the one that scored serviceability and parts logistics alongside price, and the dealer who can document repair times and battery behaviour is the one who wins the fleet order without discounting.