The pet-GPS company sold me ten days of battery life. I get twenty-three hours. That’s not a bug in my device. It’s a structural mismatch between a 2G-first tracker and a rural French village whose 2G coverage is being quietly disassembled. Several things had to be true at once for this to happen — Sigfox going bankrupt, Bouygues killing its LoRa network, Apple’s network being useless among vineyards. Here’s what I learned, the order I learned it in, and the four-paragraph version of my new respect for radio-physics.
The cat
His name is Jambon. He weighs about 4.5 kilos. He lives on the edge of a small village in rural Aude, France, between vineyards and a road I do not love. He goes outside for roughly six months a year. He covers, at a generous estimate, three-and-a-half kilometres in any given direction, which is too far for a name yelled from a doorway and exactly the use case a “10-day battery life pet GPS tracker” exists for.
That’s what I bought. The tracker is a perfectly good piece of hardware from a perfectly reputable French company. Its claimed battery life is ten days. Its actual battery life, on my cat, in my village, is twenty-three hours. I have now charged this thing more than three hundred times.
I was annoyed enough to investigate, and the investigation went deeper than I expected.
What I assumed (wrong)
My first assumption: the battery is degrading. LiPo cells lose capacity over time, especially when charged every two days for a year. Order a new device, problem solved.
My second assumption: a firmware update broke something. The “10 days” promise must be from an older firmware. Update, problem solved.
My third assumption: my cat is somehow special. He’s an outdoor cat in a rural area. Maybe he’s just outside the device’s comfort zone.
The first two are wrong. The third is almost right, but for a reason that has nothing to do with the cat.
What’s actually happening
The tracker has a multi-band module: 2G, LTE-M, NB-IoT. Its firmware prefers 2G when it can find it — 2G has the best historical coverage in Europe and works on every operator’s network. In a city, this is a reasonable default.
In my village, it is a disaster. Here’s why.
Pexiora, rural Aude, France — what the tracker sees
─────────────────────────────────────────────────────
2G ▌▌░░░░░░░░░░░░░░░░░░ ← intermittent, weak
LTE-M ▌▌▌▌▌▌▌▌▌▌▌▌▌▌▌▌░░░ ← strong (Orange has a local antenna)
NB-IoT ▌▌▌▌▌▌▌░░░░░░░░░░░ ← present but high-latencyThe 2G signal is just strong enough that the tracker tries to use it, and just weak enough that it fails to maintain the connection. Every failed connection attempt costs energy — roughly 10 minutes of normal operation per attempt. The device spends its battery looking for a network it can’t reliably hold.
The tracker manufacturer knows this. Buried in their help centre is the sentence: “battery life can be significantly reduced, potentially lasting less than 24 hours in poor network conditions.” That’s not a hedge. That’s a confession. My cat is wearing a device whose battery problem is documented as a feature of the architecture.
The 2G-preference isn’t optimisable in firmware. It’s how the radio stack is wired. Even if the manufacturer wanted to fix it, “switch to LTE-M as the default” is a chipset-level decision they made years ago, and the device on my cat’s collar is the device that made it.
The European context, briefly
This wouldn’t have been a story ten years ago. Three things changed:
2G is being decommissioned, country by country. Most European 2G shutdowns are scheduled between 2025 and 2028. The coverage map is already pocked with holes in rural areas because operators have stopped maintaining 2G antennas while they wait for the shutdown date. My village’s 2G coverage in 2026 is materially worse than it was in 2022.
Sigfox went into insolvency in 2025. The low-power network that was supposed to be the cheap, long-range alternative for IoT devices is in administration. There are no consumer trackers being shipped that use it anymore. The ones that exist are operating on a network that may not be there next year.
Bouygues killed LoRa for consumers in December 2024. The French LoRa operator Objenious — which several pet-tracker companies depended on — was shut down. Any tracker that used that network is now a brick. There’s a generation of devices on shelves and in attics that quietly stopped working when the network died.
The IoT-network landscape that existed when the trackers in my drawer were designed is not the landscape they operate in now. The hardware bet on a future that didn’t arrive.
The four options for trackers, ranked by how reliable they’re not
I made a chart. The chart is depressing.
| Option | What it depends on | What’s going wrong |
|---|---|---|
| 2G-first cellular | A network being phased out | What I have now; dies in 23h |
| LoRa via consumer network | Bouygues Objenious (dead) or TTN (no rural gateways) | The network died; or there’s no gateway within 150km |
| Sigfox | A company in administration | Likely dead by next year |
| Apple Find My / Google Find Hub | Other people’s phones being nearby | Useless in a vineyard where nobody walks |
| LTE-M-first cellular | An actively-deployed network with rural coverage | This is the one that still works |
| Hybrid LTE-M + community network | Both of the above | Best fallback |
Apple Find My is the one that surprised me. AirTags are excellent in cities — they crowd-source location off the iPhones of everyone walking past. In a rural village, there are no iPhones walking past. The vineyard is silent, density-wise. An AirTag on a roaming cat would report a position every several hours from several hundred metres away. Useless is the technical term.
What survives is LTE-M. It’s the one cellular technology that’s being actively deployed for low-power devices, has real rural penetration (lower frequency band, better at going through walls and trees), and is supported on the major French operator’s network at my address. The devices that use it as a first choice — not a fallback after 2G — actually deliver multiple days of battery life in my conditions.
The discovery was buried in a tracker spec sheet: “LTE-M Cat-M1 multi-operator, with 2G as fallback.” That word order — LTE-M first, 2G as fallback — is the single sentence that turns a 23-hour battery into a 3-5 day battery, on the same cat, in the same village, with the same radio environment.
What I bought, and why
I replaced the device with a current-generation LTE-M-first tracker (the specific brand is in the larger note; the relevant fact is the first). Battery on the same cat, same village, same usage pattern: about four days. That’s not the ten days the marketing department promised, but it’s an order of magnitude better than what I had.
The lesson, narrowly: when the network you bought is being decommissioned, your problem is the network, not the device.
The lesson, broadly: a piece of hardware sold for a use case in 2022 may not be a piece of hardware that works for that use case in 2026, even if the device itself is exactly the same. The world around it changed. The radio spectrum changed. The economics of supporting unprofitable networks changed. Marketing copy did not change.
Three things I didn’t expect to learn
Pet GPS is a real engineering field. I thought it was a packaging problem — slap a GPS chip in a collar, ship it. It isn’t. The battery / antenna / radio stack / network preference / firmware update path is a stack of compromises, and where each manufacturer compromises tells you what year they designed the device. The new entrants are LTE-M-first. The incumbents are 2G-first because that was the right call when they started shipping.
Marketing copy uses “battery life” the way it uses “miles per gallon.” It’s a real number, measured under conditions designed to maximise it. The disclosure is in the help centre. Read the help centre before you buy the device. (I know. I know.)
Rural infrastructure is invisible until it isn’t. The Bouygues Objenious shutdown affected, conservatively, every IoT device in France that had been sold on “LoRa network coverage included for 3 years.” Those people are now holding bricks. The shutdown got essentially no press in the consumer-tech sphere. The bricks are silent. If you build anything that depends on a low-power-WAN network in 2026, which network and who is paying to keep it alive is a more important question than which device you put on top of it.
The pragmatic version
If you’re reading this because your tracker is also dying in a day: check whether it’s LTE-M-first or 2G-first. If it’s 2G-first and you’re in a rural area, no firmware update will save you. The device is shipping a 2022 default into a 2026 network landscape. Get a current-generation LTE-M-first tracker and add a tile/airtag-class community-network device as a backup for the case where the cellular battery runs out. Total cost: less than a vet visit. Sleep improvement: real.
If you’re not currently dealing with this: enjoy the moment. It is a delight to be reading this and not investigating it. The cat is, as of this writing, outside.
The cat is fine. The tracker is fine. The infrastructure is, slowly, fine.