Answer first: a first pilot should usually include more than one station class, but the mix must be small enough to compare. Use compact 4- or 6-slot stations to test low-commitment venues and visibility, 8- or 12-slot stations for proven moderate demand, and a limited number of 16- or 24-slot-class stations for measured high-peak locations. Match sizes within comparable venue groups, define success before launch, and reserve the right to relocate rather than treating the initial mix as a fleet forecast.
A pilot that buys only the smallest model cannot show whether failures came from weak demand or insufficient capacity. A pilot that buys only large cabinets may hide return or replenishment problems under excess inventory and makes relocation harder. A mixed portfolio reveals more, provided the test is designed to separate station size from venue quality.
The objective is not to represent every possible product. It is to answer a few launch questions: Which venue types create completed and supportable rentals? What borrow and return buffer is needed through the local peak? How often must the operator visit? Can the hardware, software, payment, and field process recover from faults? Which station class is easiest for the venue to accept and keep visible?
Create three capacity roles
Define roles rather than promising exact performance:
- Compact discovery role: 4- or 6-slot stations for counters, reception points, small hospitality, salons, studios, or low-density retail. The goal is to test visibility, payment conversion, basic servicing, and venue acceptance.
- Moderate service role: 8- or 12-slot stations for established restaurants, hotels, gyms, entertainment areas, or malls where demand has a measurable rhythm and more borrow/return buffer is useful.
- Peak resilience role: a limited 16- or 24-slot class for a transport zone, hospital public lobby, convention area, event, or other site with a measured wave, slower service access, and high support impact if the station becomes empty or full.
These are pilot labels, not universal venue mappings. A busy small restaurant may need more buffer than a spacious but quiet lobby. The pilot must retain station-level evidence.
Avoid confounding size with venue quality
If every compact station goes to hidden, low-traffic venues and every large station goes to premium entrances, the pilot cannot reveal the effect of capacity. NIST's engineering statistics guidance explains blocking as a way to control nuisance factors while comparing a factor of interest [1]. A commercial pilot will rarely be a perfect experiment, but the principle is useful.
Group comparable venues by attributes such as category, visible placement, opening hours, payment flow, neighborhood, and expected peak. Within a group, test more than one capacity or initial loading strategy where practical. Alternatively, rotate stations between approved comparable locations after a stable period. Record changes in signage, pricing, promotions, staff behavior, software versions, and weather or events that could distort the result.
Do not over-randomize a live operation. Venue safety, contracts, electrical access, and customer clarity come first. The purpose is a fairer comparison, not academic purity.
A practical first-city allocation method
Begin with venue hypotheses, not a fixed purchasing ratio. For example, select several compact sites to test acquisition and density, several medium sites to test routine service, and one or two high-peak sites only where access and demand evidence are credible. Keep spare units or swap capacity so an obviously mismatched model can be replaced without waiting for a new shipment.
The number of locations depends on budget, geography, field capacity, and statistical uncertainty. No source supports a universal “60/30/10” mix. A small but well-instrumented pilot is more useful than a wide rollout with missing timestamps, unclear station ownership, and inconsistent payment setup.
Government guidance on testing and piloting services recommends defining objectives, scope, scale, realistic conditions, evaluation methods, operational assumptions, system capability, and next actions before rollout [2]. Apply that discipline to the station portfolio: write what each class is meant to learn and what evidence will trigger keep, resize, relocate, or stop.
Use one measurement contract
All station classes should report the same core events and states: scans, payment outcomes, dispense acknowledgement, return recognition, ready inventory, working empty slots, disabled bays, offline time, support contacts, and field interventions. GBFS's uniform real-time status model offers a useful analogy for consistent station reporting [3]. A mixed pilot is hard to compare if each hardware model emits different identifiers or state definitions.
Normalize timestamps, time zones, venue hours, and observation windows. Review 15-minute peaks as well as daily totals. Include failed journeys and time below service thresholds. Compare field workload per station and per completed rental without publishing a projected ROI from a small sample.
Define decision gates before launch
For every station, set a review date and four possible decisions:
- Keep: service buffers, venue acceptance, and operational workload meet the pilot objective.
- Resize: repeated borrow or return failures are attributable to capacity.
- Relocate: demand remains weak despite visibility, uptime, inventory, and functioning payment.
- Repair or reintegrate: hardware, connectivity, mapping, or order-state defects prevent a valid capacity conclusion.
Require enough representative weekdays, weekends, and venue-specific peaks before deciding. Seasonal or event venues need matched periods. Document exclusions rather than deleting inconvenient data.
Plan the next purchase from roles, not winners
The pilot may show that compact stations are useful for density, medium stations for stable routine sites, and large cabinets for a small number of peaks. That does not produce one winning model. It produces a portfolio policy: which role each station serves, how it is initially loaded, when it is visited, and when it moves.
The most valuable outcome is not a confident-looking fleet percentage. It is a reusable decision system grounded in comparable data. A mixed pilot should reduce uncertainty about venue fit, borrow and return capacity, field workload, and system compatibility before the buyer commits to scale.
Evidence date and limits
Evidence reviewed through 2026-08-09. The following limits are part of this buyer guide:
- Capacity roles and station classes are proposed pilot constructs, not standards or guaranteed venue mappings.
- Experimental-design guidance is adapted conservatively to live commercial operations; it does not make the pilot a controlled scientific trial.
- Local venue, safety, accessibility, electrical, payment, privacy, product, import, and consumer rules require review. Verify target-country requirements before publication and deployment.
- No sample size, conversion, ROI, expansion ratio, or customer outcome is guaranteed.
Related CoreCharge guides
- shared power bank business starter guide
- choose shared power bank stations from china
- products 4 slot.html
- products 8 slot.html
- products 12 slot.html
- products 24 slot.html
- Previous guide in this capacity cluster
- Next guide in this capacity cluster
Sources
1. NIST/SEMATECH - e-Handbook of Statistical Methods: Randomized Block Designs (accessed 2026-08-09). blocking can reduce the influence of nuisance factors when comparing a primary factor; adapted here as a pilot-design principle. 2. UK Government - Testing and Piloting Services Guidance Note (accessed 2026-08-09). pilots should define objectives, realistic scope and scale, evaluation methods, operational assumptions, system capability, and next actions. 3. MobilityData - General Bikeshare Feed Specification (GBFS) (accessed 2026-08-09). uniform station-status definitions make shared-mobility availability easier to compare across a system.