Netflix recommends a show based on your viewing history. You start watching. Three episodes in, you’re not really into it. You finish it anyway because you’ve already invested time.

A friend recommends a show during coffee. You start watching. You love it from the first episode.

What’s the difference?

According to a Nielsen study, 83% of people trust recommendations from people they know, while only a small percentage trust algorithmic suggestions the same way. That gap is interesting. Because algorithms know way more about your viewing habits than your friends do.

So why do algorithms underperform?

OK here’s what I’ve been thinking about. The algorithm knows what you’ve watched. It doesn’t know why you watched it.

You finished that documentary series because it was on during a long weekend. The algorithm interprets that as deep interest. Actually, you just didn’t feel like choosing anything new. Now it’s feeding you more documentaries when you really wanted a comedy.

What the algorithm sees	What your friend sees
You watched X for 8 hours	You complained about work last week
You gave it thumbs up	You mentioned wanting something funny
You finished the series	You said you hated the last recommendation
Similar users liked it	Your specific taste in dark humor

The algorithm sees behavior. Your friend sees you.

The mood problem

Right, so here’s another thing. Your taste isn’t fixed. It shifts with your mood, your week, your energy.

Tuesday you want something light. Saturday you want something intense. Sunday you want something comforting. The algorithm doesn’t know which Tuesday you’re having.

Your friend asks: “How’s your week going?” and recommends accordingly. That context matters more than any viewing history data.

Sophie described it well: “Netflix keeps recommending me crime thrillers because I watched one during a specific stressful period. I don’t even like them usually. My sister knows I actually want romantic comedies right now.”

Why the algorithm keeps getting it wrong

It optimizes for engagement, not satisfaction. This is the big one. Netflix wants you watching more, not necessarily enjoying what you watch more. These aren’t the same thing.

A recommendation that hooks you and keeps you binging even when you’re not loving it? That’s a win for the algorithm. But it’s not really a win for you.

It can’t tell why you abandoned something. Did you quit because it was bad? Because you weren’t in the mood? Because life got busy? The algorithm just sees “not finished” and adjusts accordingly, but probably wrong.

Similar users aren’t you. Just because people who watched X also watched Y doesn’t mean you will. That statistical similarity misses everything that makes taste personal.

What friends have that algorithms don’t

They know your life. They know you just broke up, had a hard week, are in a creative phase, hate your job. That context shapes what’s actually right for you right now.

They’ve absorbed your feedback. Every time you’ve said “I loved that” or “I couldn’t get into it,” they’ve updated their mental model of you. And they can understand nuance. “You’ll like the first three seasons, skip four, come back for five.”

They filter for quality. Your friend isn’t recommending something they hate. The algorithm might recommend things just because similar users watched them. Different standards.

They can describe the vibe. “It’s like X but with better characters” tells you way more than a Netflix thumbnail. You can decide if you’re in the mood for that vibe.

The cost of algorithm-only discovery

When you only get recommendations from Netflix’s algorithm, things narrow:

You watch more of what you’ve already watched. The algorithm reinforces your existing patterns. You discover less new stuff outside your established preferences. Your taste actually calcifies.

Over time, you end up in a bubble of familiar-feeling content. Nothing surprising. Nothing outside your usual lane.

This is what streaming fatigue often actually is. Not too many shows. Too many shows that all feel vaguely the same because they’re coming from the same recommendation system.

How to get better recommendations

Ask people directly. Not “anyone have show recs?” on social media. That gets you the same trending list everyone else is getting. Ask specific people with similar taste: “Hey, you know I liked X. What should I watch next?”

Describe your mood when asking. “I want something funny but not stupid” or “I need a drama that won’t depress me” gives people something to work with.

Keep a received recommendations list. Not everything sounds good immediately. Write them down. When you’re stuck for what to watch, pull from there instead of scrolling.

Reciprocate. If you want recommendations, give them. The people who recommend best are people who think about what you’d like. That relationship takes cultivation.

The awkward part: acting on recommendations

OK one thing I’ll admit. I’m not great at actually watching what people recommend to me. I nod, I say “I’ll check that out,” and then I go back to scrolling.

If that’s you too, here’s what might help. Add it to your list immediately. Not “I’ll remember.” Open your phone, write it down. Next time you can’t decide what to watch, check the list first.

Also, try the first episode the same week you got the recommendation. The enthusiasm from the conversation helps. A month later, you’ve forgotten why it sounded good.

Jake started doing this and realized he was watching way better stuff. “I used to scroll for 20 minutes and settle for whatever. Now I pull from the list and I’m actually interested from the start.”

Use Streaming Video Pause to break out of algorithm loops

One side effect of the autoplay system: you rarely pause to actually think about what you’re watching. You just keep going because the next episode starts.

The pause between episodes is a good moment to ask: “Am I actually enjoying this, or am I just continuing?” If it’s the second, maybe check your recommendations list instead.

That small break creates room for intentional choice. Which matters when algorithms are designed to keep you in default mode.

When algorithms actually do help

I don’t want to pretend algorithms are useless. They’re fine for:

Finding things in genres you already like. Discovering new shows in your established comfort zone. Surfacing specific things you might have missed.

But for genuine discovery, taste expansion, or recommendations that actually fit your current life? Humans win.

The mix that works: algorithms for maintenance, friends for discovery.

FAQ

What if my friends have bad taste?

Then they’re not your recommendation friends. You don’t need all friends to recommend shows. Find the one or two whose taste aligns with yours (or expands yours in directions you trust) and lean on them.

I don’t have friends who watch what I like.

Online communities help here. Reddit threads for specific genres, Discord servers, book clubs that sometimes discuss adaptations. You can find people with aligned taste even if no one in your immediate life shares it.

Is there any way to make algorithms give better recommendations?

Somewhat. Actively rate things. Be honest about what you didn’t like. Watch a wider variety deliberately. But honestly, the improvements are marginal compared to just asking someone who knows you.

The algorithm is optimizing for Netflix’s goals, not yours. Your friends, when they’re thoughtful about it, can beat that pretty easily. Not because they have more data, but because they have context. Use that. Ask people whose taste you trust. Keep a list. And when in doubt, pick from the list instead of scrolling.