Start with what a device is built to do
What a connected device collects depends mostly on its job. A leak detector measures moisture. A wearable measures movement and pulse. A smart speaker listens for commands. But collection rarely stops at the sensor. A device also generates information about itself, its owner’s account and the way it is used. Looking at the categories one at a time helps you see what is really being gathered.
Common categories of data
Measurements from sensors
This is the core data of most devices: temperature, humidity, light, motion, vibration, air quality, energy use, and so on. The types of sensors involved determine the readings. Taken alone, a single reading seems harmless. Collected over weeks, it can show routines, such as when rooms are occupied.
Usage and activity data
Records of how you use a product: when a light was switched on, which settings you changed, what programs or routines run, how often an app is opened, which features you tap. Makers use this to see which features matter and where people get stuck. Usage history from a smart plug or appliance can also hint at daily habits.
Account and identity details
Names, email addresses, phone numbers, payment details, delivery addresses and sometimes birthdates or photos. These come from registration and from purchasing. Accounts link a device to a person, which makes other data easier to attribute to them.
Device and technical data
Information about the hardware and its connection: model, serial number, software version, battery level, error logs, signal strength, network name and addresses. This data is important for diagnosing faults and delivering updates, and it also acts as a persistent identifier for the unit.
Location
Location may come from satellite positioning, nearby Wi-Fi networks, cell towers, or simply the address you entered. Location trackers and wearables depend on it, but many other products ask for it without needing it much. Precise location over time reveals home, work and travel patterns, so it is among the more sensitive categories.
Audio and voice
Microphones capture commands, and sometimes background sound. Systems with voice assistants typically listen for a wake word and then send the following speech for processing. Recordings and text transcripts may be stored, depending on the product and settings. Audio can include other people’s voices and private conversations.
Images and video
Cameras and image sensors capture faces, rooms, visitors and vehicles. Some devices analyze images on the unit and send only results, such as “person detected.” Others upload footage for storage and analysis.
Health and body data
Wearables and home medical devices may collect heart rate, sleep, steps, body temperature, blood glucose or similar signals. These are considered sensitive in many frameworks, because they say something about a person’s body and condition. The wearable privacy of these readings is an area of particular concern.
Behavioral and inferred data
Companies may combine the categories above to draw conclusions: when you are home, what your sleep pattern looks like, which products you might buy, or whether your routine changed. Inferences may be wrong, and they may not appear in the settings you can see.
Data about other people
Guests, children, neighbors and passersby may be recorded by a device they never chose. This is a separate category in practice, and one reason why consent in shared spaces is difficult.
Why makers collect data
Reasons are usually a mix of the useful and the commercial.
- To make the product work. A thermostat has to know the temperature, and a voice command has to be understood.
- To support the product. Logs and diagnostics help fix problems and deliver updates.
- To improve it. Analysis of usage helps designers refine features, and large pools of data can be used to train machine learning models.
- To personalize. Suggestions, schedules and alerts can adapt to you.
- To secure it. Monitoring can spot abuse or unusual behavior.
- To earn revenue. Data can support advertising, partnerships, subscription tiers or the sale of aggregated insights.
- To meet obligations. Records may be kept for accounting, safety or legal reasons.
The first few reasons are mostly tied to what the buyer wants. The last ones are more about the business, which is why it matters to read what a company says it does with the data.
Where the data goes
Collected readings may stay on the device, move to a hub in the home, or travel to cloud servers. Processing near the source, as in edge computing, can reduce how much leaves the home. Data can also pass to service providers, such as storage or analytics firms, and to partners. Each hop adds a copy and another set of rules.
Needed versus nice-to-have
A helpful test for any data point is whether the product could do its stated job without it. A light that wants access to your contacts, or a simple sensor that needs precise location, is asking for more than its function seems to require. Excess collection raises the stakes if the data is breached or reused. Likewise, how long it is kept affects the risk, a point covered further under data retention.
Seeing and controlling what is collected
- Read the privacy summary or data section in the product’s app and website.
- Check app permissions on your phone for location, microphone, camera, contacts and Bluetooth.
- Look for a dashboard that shows stored history, and for options to export or delete it.
- Turn off optional analytics, personalization and sharing.
- Choose devices with local-only modes when you can.
Policies vary widely and change over time, so the exact data any given device collects is best confirmed from its own documentation. The aim is not to avoid all collection, which is impossible, but to know what you are trading for each feature.