Algorithms decide what to show. Auctions decide who pays to be seen. But none of this would be possible without the third force: analysis.
Analysis is what makes the other two work. It is the engine underneath — the continuous process of collecting data, finding patterns in it, and using those patterns to make predictions about what a person will do next.
The internet does not just observe your child. It studies them.
What gets collected?
Every interaction a child has online leaves a trace. Most of these traces are collected automatically, without any deliberate action on the child's part.
What they click
Every tap, every link, every button
What they watch
How long, how often, at what point they stop
What they search for
The exact words they type
When they are online
Time of day, day of week, duration of sessions
What device they use
Phone, tablet, games console, laptop
Where they are
Location data, where available
What they scroll past
Even content they do not engage with
Who they interact with
Likes, comments, shares, follows
Individually, none of these data points seem significant. Together, they form a detailed behavioural profile — one that can be surprisingly accurate about who a person is, what they want, and what they might do next.
What is done with that data?
The data is not just stored. It is analysed — fed into systems that look for patterns and use those patterns to make predictions.
Observed: A child who watches cooking videos at 4pm on weekdays
Predicted: is predicted to be at home after school, likely female, probably aged 10–14, interested in food and home life.
Observed: A child who searches for football scores every Saturday morning
Predicted: is predicted to be male, sports interested, possibly a regular match watcher — a valuable audience for sportswear and gaming brands.
Observed: A child who watches anxiety or mental health content late at night
Predicted: is flagged as emotionally vulnerable at a certain time of day — which affects what content and what tone of advertising they are shown.
Platforms do not need to know who your child is. They only need to know what your child does — and from that, they can make remarkably accurate guesses.
How a profile is built
Data analysis does not stop at one platform. Many companies share or sell data signals, meaning a profile can be built from behaviour across multiple apps, websites and devices.
A child who plays a free game, watches YouTube, visits a shopping site and uses a social app may be contributing data to the same underlying profile — without ever signing up for anything or giving their name.
This is sometimes called a data shadow.
It follows a person across the internet. It updates continuously. And it is used to decide what to show, what to sell, and how to keep attention for longer.
Children are particularly valuable to advertisers because their preferences are still forming — meaning early exposure to brands and content can shape tastes for years.
What parents can do with this knowledge
You cannot opt a child entirely out of data collection online. But you can help them understand that it is happening — and that their behaviour is being watched, interpreted and acted upon.
Useful questions to start at home:
Do you think this app knows a lot about you?
What do you think it has learned from watching what you do?
Why do you think the same advert keeps following you around?
What information do you think you have given away without realising?
If an app could describe you from your data, what do you think it would say?
Practical steps worth taking:
Review privacy settings on apps and devices together.
Turn off personalised advertising where the option exists.
Use private browsing for sensitive searches.
Be cautious with free apps — if you are not paying, your data is often the product.
Talk about why some content feels surprisingly personal.
Parent takeaway
Data analysis is what ties algorithms and auctions together. It is what makes the internet feel personal — because in a technical sense, it is.
Helping children see that they're being observed — and that their data has value — is genuinely useful.