Smart Alerts
Safety alerts only work if you still read them in week six, which is why this feature is built around subtraction rather than addition. Everything that happens is ranked by how much it actually matters, repeated events are folded into a single line, and the things that are simply normal never buzz at all. The goal is a screen that is usually quiet, so that the day it is not, you look.
Safety alerts ranked by how much they matter
Three bands, plainly labelled, so the important thing is at the top rather than wherever the clock happened to put it.
A flat list sorted by time is the wrong shape for this problem. It puts a permission being switched off underneath four ordinary arrivals at school, purely because the arrivals happened later. The version that respects your attention sorts by consequence first and time second, which is what this screen does.
The top band is act now. It holds the small number of things that need a decision or a fix: protection switched off, a permission revoked, an app that has stopped reporting, a pending request from your child. Each of these carries its action inline, so approving extra time or repairing a permission happens in the list rather than three screens away.
The middle band is worth knowing. A blocked site attempt, a new app installed, an arrival somewhere unusual. Nothing here demands an immediate response, but it is the material a Sunday evening conversation is made of. The bottom band is for the record: routine arrivals and departures, ordinary limit events, and anything you have asked to keep but not be told about.
Underneath all three sits the full history, newest first and searchable by child and by type. Muting something never deletes it, so a category you have silenced still builds a record you can read on purpose rather than being interrupted by.
Why four attempts at one site is a single line
The most important engineering in this feature is the part that stops it talking.
A child who hits a blocked page rarely hits it once. They tap it, get the block screen, go back, tap it again, try a slightly different address, and try once more. That is four or five events in ninety seconds, and a naive product would send you four or five notifications about a single moment of curiosity. By Thursday you would have muted the whole feature.
So repeats collapse. Events of the same type, for the same child, about the same thing, inside a short window, become one entry with a count and a time range: blocked attempts at one site, four times, between 16:12 and 16:15. That single line carries strictly more information than four separate lines did, because the count itself is the interesting part. One attempt is a mistyped address. Nine is a decision.
The same logic runs across types. Several place crossings within a few minutes arrive together. A batch of app installs during a Saturday afternoon setup becomes one summary rather than eleven separate messages. Where grouping would hide something urgent it does not apply, so a permission being revoked is never quietly folded into a summary about something else.
Grouping also has a quieter benefit: it makes the history readable weeks later. Scrolling back through a month of grouped entries tells you a story. Scrolling back through a month of raw events tells you nothing at all, because nobody reads four hundred lines.
The five kinds of alert, and what each one really means
Place, blocked site, keyword, new app, and permission. Everything else is a variation on one of these.
Place alerts come from the circles you drew: arrived at school, left home. These are the ones most families keep switched on and they are also the ones most likely to become noise, because a tight radius plus a wobbling indoor position produces crossings all afternoon. Widening the circle fixes more alert fatigue than muting ever will.
Blocked site alerts tell you that an address matching a category you switched on was requested and stopped. Worth being precise here: the alert says an attempt happened and names the category. It does not contain the contents of the page, because the page was never loaded, and there is no reading to report.
Keyword alerts are the narrowest of the five and the most often misunderstood. You choose a small list of words, and if one appears in a notification that lands on the phone, you are told that a word from your list matched, in which app, and when. You are not shown the message. We do not store conversations, we do not open apps to read them, and there is no transcript anywhere in this product. The alert exists to prompt a conversation, not to replace one.
New app alerts fire when something is installed, with the name and where it came from, which is how a VPN app or a lookalike social app becomes visible on the day rather than in a month. Permission alerts are the housekeeping ones: location switched off, the app force stopped, protection disabled. They are unglamorous and they are the most important category on the list, because everything else depends on them being true.
Muting the safety alerts you do not need, without going blind
Two controls, per child and per type, plus quiet hours, and one rule about what you should never mute.
Every alert type has two independent settings: whether it pushes to your phone, and whether it appears in the stream. Most families end up pushing place and permission alerts, keeping blocked sites and new apps in the stream to review when they open the app, and turning the routine band off entirely. That takes about a minute to set and it is the single highest-value minute on this page.
Quiet hours cover the rest. Alerts that arrive overnight can be held until morning without being lost, which means you can keep a category switched on without agreeing to be woken by it. A held alert keeps its original timestamp, so a 01:40 event still reads as 01:40 when you see it at seven.
Settings are per child, which matters more as ages spread out. A nine year old might justify arrival alerts at three places. A sixteen year old with the same settings would generate noise about a life that is, appropriately, no longer minutely supervised. Reducing alerts as a child gets older is a feature of the relationship, not a lapse in the tool.
The one thing worth never muting is the permission and protection category. If location has been switched off or the app has been stopped, every other alert on this page silently becomes untrue, and a screen that is quiet because nothing is happening should never be confused with a screen that is quiet because nothing is being reported.
Alert fatigue, and why quiet safety alerts get read
Every design decision here trades coverage for credibility, on purpose.
Alert fatigue is not a personal failing, it is a well documented property of human attention. Hospitals learned it with monitor alarms, drivers learned it with dashboard warnings, and anyone who has installed a security camera app has learned it with motion notifications. Past a certain rate, notifications stop being read and start being dismissed as a reflex. The person is not careless. The system is badly designed.
A parental control app that buzzes forty times a day has effectively been switched off, with the extra cruelty that the parent believes it is still working. That is worse than no alerts at all, because it produces false confidence. So we would rather send you eight useful messages a week than eighty accurate ones, and we accept that this means occasionally not telling you something small.
That is what ranking, grouping, quiet hours and per-type muting are all for. Each one exists to reduce volume without reducing the signal. The measure of success is not how much this app tells you. It is whether, on the day something genuinely matters, your thumb still stops on the notification instead of swiping it away.
It follows that a quiet week is the normal outcome and the intended one. No alerts does not mean the app is broken. It means the phone checked in on schedule, permissions held, no blocked category was requested, and everybody went where they said they were going. If you want to confirm that positively rather than infer it, the device health page will show you the last check-in time and the state of every permission.
Choose what pushes and what waits
Set each alert type twice: push to your phone, or stream only. Place and permission alerts usually push, everything else can wait until you open the app.
- Settings are per child and per type.
- Quiet hours hold overnight alerts until morning.
- Muting never deletes the history.
Read the day in one screen
The today view puts the act now band at the top with its actions attached, then everything worth knowing, then the routine record underneath.
- Approve a request without leaving the list.
- Grouped repeats show a count and a time range.
- Nothing normal generates a line.
Fix the source, not the symptom
Most noise comes from one place drawn too tight. Widen the circle by thirty metres and a category that produced twenty alerts a week produces two.
- Check the noisiest place first.
- Turn off routine arrivals you never read.
- Never mute permission alerts.
Act now, worth knowing, and for the record, with counts on anything repeated.
Same type, same child, same subject, short window: one line with a count.
No message contents, no transcripts, no scoring, no alert your child cannot see.
What your child sees
Every alert type that is switched on appears in your child’s app, including the keyword list itself. When an alert is sent, their app records that it was sent and what kind it was. A family that has agreed which events raise a flag is in a completely different position from one where a child eventually discovers a hidden trigger, and that difference is the whole argument for building it this way.
- The list is not secretWhich alerts are on, and the watched words themselves, are visible to them.
- They see it was sentEach alert is recorded on their side as well as yours.
- No contents, everA keyword alert says a word matched. It never carries the message.
| Alert type | What it tells you | Band | Suggested setting |
|---|---|---|---|
| Permission or protection off | Location, app or protection has been disabled | Act now | Push, never mute |
| Request from your child | Extra time, or an app exception, with their reason | Act now | Push |
| Keyword match | A word from your list appeared in a notification | Act now | Push, with a short list |
| New app installed | App name and where it came from | Worth knowing | Stream only |
| Blocked site attempt | Category matched, with a count if repeated | Worth knowing | Stream only |
| Place arrival or departure | Named place, with the time of the crossing | Varies by place | Push for school, stream for the rest |
No. Ranking, grouping and quiet hours exist specifically to prevent it, and most families settle at well under ten pushed alerts a week per child.
No. It tells you a word from your list matched, in which app, and when. Message contents are never read, stored, or sent to you.
No. Muting stops the interruption only. Everything still lands in the history where you can read it when you choose to.
Usually because nothing has happened, which is the intended outcome. To confirm it positively, check the device health page for the last check-in and permission states.
Widen its radius by twenty or thirty metres rather than muting it. Indoor position drift across a tight circle is the cause of most repeated crossings.
Yes. Every type is set per child, so a younger child can push three categories while an older one pushes only permissions and requests.
Set your alerts once and let the screen go quiet
Push permissions and one place, leave the rest in the stream, add quiet hours, and let a mostly empty screen be the good news it is.
Related features: Places and Geofencing, Web Filtering, and Device Health.