Half your SIMs went over their allowance last month. The other half used a fraction of theirs. You paid for both.
That is the arithmetic of per-SIM data plans on a fleet where usage varies, and it gets worse as the fleet grows. Data pooling changes it by treating the fleet's allowance as one number rather than a thousand.
A shared data pool combines the allocation of every SIM in the fleet into a single allowance. A device having a heavy month draws on the headroom of devices having a quiet one. Nothing is stranded on a SIM that will never use it, and nothing tips into an overage charge while unused data sits next door.
The devices know nothing about it. Pooling is a commercial and management construct, not a change to how a device connects.
2 failure modes, and most fleets run both at once.
Overage is the visible one. A SIM crosses its allowance and the excess is billed at a rate nobody negotiated with the same attention as the base plan. One firmware rollout, one chatty device, one integration that started polling every 30 seconds instead of every 5 minutes, and the invoice moves.
Stranded data is the invisible one. A device that reports twice a day uses a small share of what you bought for it. Multiply the unused remainder across several thousand SIMs and the number stops being small. It never appears as a line on the invoice, because you were charged for it up front.
The reflex fix is to size every plan for the worst month and accept the waste. That holds until finance asks why the connectivity line grew faster than the fleet.
Pool size is average usage per SIM multiplied by the number of active SIMs. The arithmetic is straightforward. Getting the average right is the work.
Base it on measured usage, not the datasheet. A device specified at 5MB a month behaves differently once it is pulling OTA updates, retrying failed sessions on a weak signal, and carrying diagnostic payloads nobody planned for. Retries alone move the number on fleets deployed at the edge of coverage.
Then plan for growth. Pool size is set at subscription and adjustable as the fleet grows, so a pool sized for today with a known path to a larger one beats a pool sized for an optimistic year 3.
This is the trade-off, and it deserves a plain answer. Per-SIM plans fail one device at a time. A pool fails all at once. If the shared allowance is exhausted, every SIM in the fleet is affected until it is topped up.
2 things follow. Set threshold alerts in the IXT CMP and watch them, because pool consumption is the number that matters rather than any individual device's usage. And treat resizing as a decision you make deliberately. IXT does not resize a pool mid-contract without a request from you, which keeps the commercial terms predictable and puts the job of watching the trend on your side.
Pooling earns its place when usage varies across the fleet. A mixed estate of gateways, sensors, and vehicle trackers has a wide distribution, and the wider the distribution the more a pool recovers.
It earns its place on uniform fleets too, for a different reason. Even where every device uses roughly the same volume, one allowance replaces thousands of individual limits to monitor, top up, and reconcile. The saving there is operational rather than financial.
Pascal Technologies runs a 5GB shared pool across its AirHull vessel fleet, which removes per-vessel overages as boats move between countries, according to the customer story published at ixt.io.
Pooling earns its place least on a small fleet with predictable, uniform, low usage. Under 100 devices with no surprises in the usage curve, the management overhead of per-SIM plans is not the problem you have.
A shared pool changes how data is consumed and billed. It does nothing else, and 3 gaps are worth naming.
It does not improve coverage. A device that fails to attach in a market where your provider holds no roaming agreement fails the same way with a pool behind it.
It does not add security. Traffic still crosses whatever path it crossed before. Keeping device traffic off the public internet is the job of a private APN, and IXT SecureNet is where that sits.
It does not reduce the volume a device sends. If consumption is growing because of a polling interval or a retry loop, a pool absorbs the cost quietly for a while and then presents it in one bill.
4 questions separate a real pool from a marketing one.
Does the pool work across borders, or does it reset per country. A pool that stops at a border is a set of national pools with one invoice.
How fresh is the consumption data. Many provider portals report usage 24 to 48 hours late. Managing a pool on day-old figures means finding out you are at 95% when you are already at 100%. TheIXT CMP shows usage in real time with threshold alerts.
What happens at exhaustion. Throttling, suspension, overage billing, and automatic top-up are 4 different answers with 4 different consequences for uptime.
How does resizing work, and on what notice. Ask before you need it.
Pooling is one decision inside a connectivity subscription, not a strategy. It belongs alongside the questions that outlast it. Which networks the SIM reaches. Whether traffic stays off the public internet. Whether you find out about a failure in real time or the following afternoon.
Ask us how it works for your deployment.