June 4, 2026 · 6 min read · Technical Whitepapers
100 devices on individual 10MB plans buying 1,000 MB total, using 700 MB — 300 MB wasted, paid for, and never used. Same devices on a pooled 1,000 MB plan: one bill, zero waste. The math is simple. The procurement trap is signing individual plans at small scale and forgetting to switch when the fleet hits 500 devices.
IoT data pooling is not a pricing trick. It is an inventory management problem solved at the network layer. When 100 devices each have a 10 MB individual plan, the fleet buys 1,000 MB per month. If 20 devices use 2 MB each (trackers in storage) and 5 devices use 50 MB each (video cameras on a highway), the fleet pays for 1,000 MB but uses 700 MB. The 300 MB of unused allocation is waste that a pooled plan eliminates — because the video cameras draw from the same pool that the idle trackers leave untouched. The mathematics of pooling works at any scale. The procurement trap is that most fleets start on individual plans at 50 devices and never switch.
| Model | How It Works | Best For | Worst For |
|-------|-------------|----------|-----------|
| Pay-as-you-go (PAYG) | Pay per MB consumed | Pilots, unpredictable usage, seasonal | Steady-state fleets (overpay vs pooled) |
| Fixed bundle per SIM | Each SIM gets fixed allowance | Uniform fleets (all same device type) | Heterogeneous fleets (waste from unused allocations) |
| Pooled / Shared | One bucket across all SIMs | Heterogeneous fleets >100 devices | <20 devices (pool overhead not worth it) |
| Hybrid | Unlimited on phones, pooled on data-only | Mixed-device enterprise fleets | Single-device-type deployments |
A real example: 100 devices, 70 use 5 MB, 20 use 15 MB, 10 use 50 MB. Total consumption: 1,150 MB. Individual 20 MB plans at EUR 0.50/SIM = EUR 50/month, but the 10 heavy devices exceed 20 MB and incur overage at EUR 1/MB. Overage adds EUR 300. Total: EUR 350. Pooled 1,500 MB plan at EUR 0.30/MB pooled = EUR 450. Pooled is EUR 100 more... in this case. Reverse the mix — 70 heavy, 30 light — and pooled wins by 40%. The model choice depends on the actual usage distribution, not the fleet size.
Source: OneSimCard IoT, "Pay-as-You-Go vs. Pooled Data Plans", 2025. Available at https://travel.onesimcard.com/m2mblog/pay-as-you-go-vs-pooled-data-plans-which-iot-sim-model-is-best-for-your-project/
The crossover point where pooled beats individual is not a fleet size. It is a usage variance threshold. When the standard deviation of monthly data consumption across devices exceeds 40% of the mean, individual plans waste more than 20% of purchased data to under-utilization. Pooled plans eliminate this waste because the pool absorbs the variance.
Three procurement rules for 2026: (1) If fleet is under 50 devices and all same device type, individual fixed bundles are simpler and often cheaper — do not overengineer. (2) If fleet is 50-500 devices with mixed types or usage patterns, run the math on both models using 3 months of actual usage data — not estimates. (3) If fleet exceeds 500 devices, pooled is almost always cheaper — the variance is large enough that individual plan waste exceeds the pooling premium.
Source: IXT, "How do global data pools reduce industrial connectivity costs?", 2025. Available at https://ixt.io/blog/how-do-global-data-pools-reduce-industrial-connectivity-costs
The highest-value form of pooling is global: one data bucket shared across devices in 30 countries. A tracker in Germany consuming 5 MB and a video camera in Brazil consuming 500 MB draw from the same pool. One invoice. One currency. No per-country carrier contracts. Global pooling requires a connectivity provider with direct MNO agreements in each country — the provider pays local rates and passes the pool through. If the provider is reselling roaming, the pool is priced at roaming markup and the economics break.
The SIM requirement: multi-IMSI with local breakout per country. The device connects at native rates on the local MNO. The data meter runs against the global pool. The provider handles the inter-operator settlement. This is not a catalogue SKU — it is a project-scoped architecture because the provider must hold agreements in every country in the deployment footprint.