You mention new sneakers to a friend over coffee. By evening, your feed is wall-to-wall shoe ads. It feels too precise to be chance — so is your phone quietly listening in?
Almost certainly not. The real explanation is less sinister, but arguably more impressive: social media platforms don’t need to eavesdrop, because your digital behaviour already says plenty about you. Every video you linger on, every post you like, every account you follow, even how long you pause mid-scroll, is a small signal. Individually, these signals mean little. Multiplied across millions of users, they form patterns an algorithm can learn from. People are creatures of habit, and that’s precisely what makes prediction possible: algorithms don’t read minds, they read routines.
That trail doesn’t stay locked inside a single app, either. Most people sign into YouTube, Gmail, Maps and Chrome with one Google account, or use a social media login to access games and shopping apps. That convenience quietly links platforms together behind the scenes. Spend an afternoon Googling trekking shoes, watch a hiking vlog on YouTube, then open Instagram, and backpack ads may appear soon after. No app is reading your thoughts. Cookies and tracking pixels, shared across advertising networks, are doing the connecting instead, and the internet is far more stitched together than it looks from the outside.
Location adds another layer. Even if you never tag a place, GPS, IP addresses and nearby Wi-Fi networks can reveal your city, neighbourhood, or regular haunts. That’s why a trip to Goa might trigger resort or watersports ads almost immediately. To an algorithm, location isn’t just geography, it’s intent.
Inside the pattern-finder
At the centre of all this sits the algorithm itself: a set of automated rules deciding what shows up in your feed next. It isn’t guessing randomly, it’s constantly comparing your behaviour against millions of others, hunting for correlations. Say a large group of space-and-rockets viewers also tends to click on sci-fi trailers. The algorithm registers that overlap and reasons, in effect, that people like this user tend to enjoy sci-fi, so it recommends more of it. The more you use a platform, the sharper these predictions get, which is exactly why a feed can start to feel unsettlingly personal. To an algorithm, you’re never just an individual scrolling alone, you’re a data point inside a much larger crowd of people with matching habits and interests, and learning from that crowd is how the system gets so good at guessing what you’ll want next.
India, by the numbers
The scale of this in India is hard to overstate. The country has crossed one billion internet subscribers, making it one of the largest sources of digital data on the planet, and Indian users spend an average of 3.2 hours a day on social media. Mobile data consumption touched roughly 229 billion gigabytes in FY2025, and a 2024 survey found that 90 percent of Indian teenagers aged 14 to 16 have smartphone access at home, with 76 percent using social media for recreation. That’s an enormous, constantly refreshed pool of behavioural data feeding these systems.
Taking back some control
Disappearing from the digital radar entirely isn’t realistic, but making yourself a harder read is possible. Auditing app permissions and regularly clearing cookies and browsing history cuts off some of the easiest data to collect. Opting out of personalised ads wherever that option exists, and simply pausing before clicking, helps too, since every tap, share and lingering scroll adds one more piece to the profile these platforms are building.
The friends who give you away
There’s a word for one part of how this plays out: homophily, from the Greek for “love of the same.” It describes our tendency to bond with people similar to us, whether in hobbies, age or opinions, and it means an algorithm doesn’t only learn from your own behaviour. If several of your friends follow photography accounts or football pages, the system may guess you’re interested too. A user who never once searches for cricket scores might still start seeing cricket content simply because most of their close contacts do, since the algorithm is reading the crowd around them as much as it’s reading them. Sometimes, it’s your friends’ clicks giving you away as much as your own.
What the research says
The research backs up just how far this goes. The Pew Research Center has found that nearly all the content users encounter on social media is selected by algorithms drawing on vast troves of personal data to maximise engagement, and the same research found that a large majority of users regularly encounter content that amuses them, while smaller but still significant shares report feeds that leave them angry or sad, a reminder of how much algorithmic curation shapes mood as well as content. A 2024 review in Psychological Bulletin concluded that everyday digital traces, from the apps people use to the accounts they follow to how they interact online, can reveal personality traits with notable accuracy. And research published in the Journal of Big Data has shown that simply analysing a user’s past behaviour lets algorithms reliably forecast what they’re likely to engage with next.
Social media doesn’t know you because it’s listening. It knows you because every search, click and pause leaves a trace, and today’s algorithms have become remarkably skilled at turning those traces into predictions. It isn’t reading your mind. It’s reading your patterns.