In late 2021, YouTube quietly removed something most of us never thought twice about: the public dislike count. The button was still there and you could still click it but the number next to it was gone.

People were not happy. And they had a good reason to be. That number wasn’t just trivia. It was a warning system. If a “5-minute fix” tutorial had a 90% dislike ratio, you knew to close the tab and save your time. The count was, in effect, the internet policing itself.

So who benefited when that warning system disappeared? Mostly big brands and advertisers, whose polished, over-produced videos had been getting quietly buried under a wall of thumbs-down.

But here’s the twist that makes this story worth telling. Researchers later found that the dislike button had been hitting one group far harder than others: female creators were receiving significantly more dislikes than men for comparable content. After the count was hidden, negative feedback dropped sharply, and women on the platform started publishing noticeably more.

So was hiding the dislike count a good move or a bad one? Depends who you ask. But here’s the real point: this wasn’t a design choice, and it wasn’t an engineering choice. No designer sat there thinking “this would look nicer without a number.” No engineer said “this is technically easier to build.” Someone, somewhere, made a decision about what the product should be. What it protects, who it favours, what story it tells the world. That’s a totally different domain, and it has a name: a product decision.

Product decisions are the invisible hand behind nearly everything you tap, swipe, and scroll through every day. Some of them are quietly brilliant. Some are, frankly, a little sinister. Once you learn to spot them, you start seeing the tech world completely differently. It becomes less like a collection of apps, and more like a series of very deliberate bets about human behaviour.

I’ve tried to break these down into four layers as follows:

Level 1: The Small Stuff That Isn’t Actually Small

Ever wondered why texting an iPhone user shows up in blue, while texting an Android user turns the bubble an ugly green with none of the nice features? Apple has some of the best designers and engineers on the planet. There’s even an industry-wide standard (called RCS) that would let Android and iPhone messages work identically. The technology to fix this exists. It’s just… not used until now.

That’s not a limitation. That’s a decision. Apple has correctly calculated that a slightly worse experience for Android users creates social pressure within friend groups and families. That nudges people toward buying an iPhone. Nobody explains this to you. You just feel it: the low-key second-class-citizen vibe of the green bubble.

TikTok’s watermark that follows your videos everywhere they get reposted? Its’ the same thing. It is not a design flourish. Rather it takes more processing power to add. It exists purely to plant TikTok’s flag on every piece of content that leaves the app. Even the little “Sent from my iPhone” signature on emails was a deliberate choice, not an oversight.

These are small, almost invisible nudges. Harmless-looking. But they all quietly serve the company, not just you.

Level 2: When Ignoring Your Users Is the Right Call

Here’s where things get genuinely counterintuitive.

Imagine you run a project management tool, and your customers are bombarding you with requests: more views, faster search, better exports. The obvious move is to line up those requests and start building. Instead, imagine a team choosing to spend their time on something nobody asked for: a tiny label on the login screen reminding you which account you used last time: Google, Apple, or email.

It sounds trivial. But people constantly forget which method they used to sign up, accidentally create a second account, and quietly disappear as a “user” without anyone noticing why. That one small addition can meaningfully cut down on this kind of silent churn.

Or think about receiving a bank verification code by text. Instead of forcing you to jump between apps. Copy a code. Paste it back. Your phone can simply notice the pattern and offer the code right above your keyboard, ready to tap. If you’d asked people back in 2017 what they wanted from their next phone update, not one of them would have probably said “please detect my verification codes automatically.” They’d have said “bigger battery” or “better camera.” Nobody asks for the thing that actually saves them the most friction, because they don’t know it’s possible yet.

This is one of the strangest and most important truths in how products get built: don’t listen to what people say they want, watch what they actually do.

It’s a genuinely brilliant principle when it’s used to remove friction you didn’t even know you had. But it has a dark side too. Instagram users say they want a simple, chronological feed of their friends’ posts. What they do is scroll for forty-five minutes through a feed curated by algorithms that has almost nothing to do with their friends. YouTube viewers say they hate Shorts. They go out of their way to install browser extensions to block them. While, on average, still watching a lot of them. If Shorts stopped keeping people on the app, they’d disappear tomorrow. Behaviour, not opinion, is what gets built for. That’s a genuinely double-edged sword.

The Facebook “People You May Know” box is maybe the most infamous example of this principle taken to its extreme. In 2007, Facebook noticed something alarming: new users who didn’t add at least seven friends in their first days almost always quit. So a team was built around solving that one number. And they solved it very well. Even if you never uploaded your contact list, if your number or email showed up in someone else’s address book, Facebook could still link it to you as a “shadow contact.” An algorithm then looked at friend-of-a-friend patterns to guess who you probably knew. This meant the average user had tens of thousands of potential matches being scored and ranked in the background.

It worked so well that a psychiatrist reportedly noticed her own patients suggesting each other as “people you may know.” Reportedly, there was even an internal rule aimed at preventing the algorithm from suggesting a man’s mistress to his wife. Whatever you think of the ethics here, it’s hard to argue the feature failed at its job. It’s also no accident that the person who led that project became one of the most influential figures in Silicon Valley.

The lesson isn’t that these decisions are automatically bad. They simply follow whatever the company has decided its priorities are. If growth is the ultimate priority, you get the Facebook friend-suggestion machine. If user wellbeing is the priority, you get YouTube quietly protecting creators by hiding a number.

Level 3: Changing What a Product Is, Without Changing What It Does

This is where product thinking stops being about small nudges and starts being about identity.

Imagine a video-feedback tool that, on the surface looks exactly like Google Drive: folders, drag-and-drop uploads, and a file browser. That’s basically what every product in that space looks like, because that’s the obvious way to build “a place to store and review videos.”

But if you actually talk to the people using these tools, you might discover something surprising: almost nobody cares about the storage. They’re only there to leave comments and feedback quickly. Meanwhile, the folder-and-drive layout, is actively getting in their way.

So imagine redesigning the whole homepage around a large, dedicated “drop a file here and give feedback immediately” section. Even though, technically, you didn’t need to build anything new; there was already an upload button elsewhere. Nothing changed under the hood. But the feeling of the product changed completely. It no longer resembles a drive, so people stop treating it like one. It becomes something else entirely: a fast feedback tool, not a storage tool. Same engineering, same features but a completely different product in the user’s mind.

Loom went through almost the exact same transformation in real life. It launched as a tool for recording yourself narrating feedback on a website during user research sessions. Which is a really niche use case with modest growth. But its team noticed people were quietly using it for something else entirely: just recording quick video messages to send to coworkers instead of writing long emails. Rather than fight that behavior, Loom leaned into it and rebranded itself as an async video-messaging tool for work. The recording technology didn’t change. The player didn’t change. But the story of what the product was for changed completely. That repositioning helped drive it to a billion-dollar acquisition a few years later.

That’s the power of a Level 3 decision: the product stays the same, but what it means to the person using it is rebuilt from scratch.

Level 4: Playing Chess With an Entire Company’s Future

At the top of the pyramid, product decisions stop being about a single feature or even a single product. They become about driving an entire portfolio of products toward one long-term goal. This is usually the actual, unglamorous day job of a CEO or founder.

Go back to Google in 2002. From the outside, the company looked unbeatable. From the inside, its founders were reportedly quite worried: nearly all web traffic ran through Internet Explorer, and if Microsoft ever decided to build a serious competing search engine and bake it into the browser everyone already used, Google’s core business could be in real trouble.

During that time, a small team led by a young product manager named Sundar Pichai was working on something that looked almost like an afterthought. A browser toolbar. On paper, it was a minor convenience feature. In reality, it was a Trojan horse. A way to keep putting Google’s search box in front of people no matter which browser they were using.

That single insight scaled into an entire strategy. Build or acquire genuinely useful products. Offering generous free email storage. A map of the entire world. A free mobile operating system. And quietly use each one as a channel back to the core business. Early Gmail nudged you to install the Google toolbar. The Google Earth installer did the same. Android, which cost billions to build, simply arrived with Google as the default search engine already baked in. And eventually, the same team that started with a browser toolbar built Google’s own browser outright. The person leading that early, unglamorous toolbar project is now the CEO of the entire company.

That’s Level 4: not a feature, not even a product, but a deliberate constellation of products all quietly pointing back at the thing that actually makes the money.

Why This Actually Matters to You

None of this is really about tech trivia. It’s about learning to read the world a little differently.

The next time an app does something slightly annoying, slightly manipulative, or surprisingly delightful. Something like a green bubble, a watermark, a feature nobody asked for that somehow makes your life easier. It’s worth taking a pause and ask: is this a limitation, or is this a choice?

Almost every time, it’s a choice. Somebody, somewhere, decided this was the version of the product that best serves the company’s goals. And sometimes at your expense, sometimes very much in your favour. Design makes things usable. Engineering makes things possible. But product decisions decide what the thing actually is, who it’s really for, and what it’s quietly trying to get you to do.

Once you start noticing that layer, you can’t really stop.