🌙 The 3am Internet

By Siri Lahari Chava

It started as a question I couldn't stop thinking about. What do people actually Google at 3am? Not what they'd admit to over coffee. What they really type in when the lights are off, everyone's asleep, and it's just them and a screen.

I had a hunch it was loneliness. Maybe doom-scrolling. Maybe people spiraling about their exes.

I was wrong.

So I built something to find out. I wrote a Python script using the Google Trends API to pull hourly search data across 20 US cities, tracking what people search between 1 and 5am, broken down by emotional category: loneliness, health anxiety, existential dread, hyperfixation, relationship pain, financial stress, escapism, and self-improvement spiraling. 57,436 data points later, here's what I found.

🩺 The Finding That Surprised Me

Health anxiety wins. Every city, every night.

Not loneliness. Not rabbit holes. At 3am, Americans are Googling their symptoms. "Is this normal." "Do I have cancer." "Heart palpitations anxiety." It's the category that consistently led across all 20 cities, beating hyperfixation, loneliness, and everything else.

There's something deeply human about that. The lights are off, the world is quiet, and your brain picks that exact moment to convince you something is very wrong with your body. You reach for your phone not to scroll, but to ask Google if you're okay.

🌆 The City That Never Really Sleeps

New York and LA show 2–3x the late-night search intensity of most other cities. That tracks. But the city that surprised me was Atlanta. consistently high across multiple emotional categories, not just one. Atlanta at 3am is having a lot of feelings.

And Dallas, my city, sits solidly mid-pack. Calm. Unbothered. Make of that what you will.

📅 The Sunday Scaries Are Real (But the Data Says Friday)

Everyone talks about the Sunday Scaries, that creeping dread that sets in Sunday evening. The data backs it up, but with a twist: the actual peak of late-night search activity isn't Sunday night. It's Friday night bleeding into 5am Saturday morning. No work tomorrow. Nowhere to be. The internet stays open.

💻 Explore the Dashboard Yourself

I built an interactive Tableau dashboard so you can dig into the data yourself, filter by city, by emotional category, by hour. See what your city is actually searching for at 3am.

🔧 How I Built It

The whole project runs on Python and Tableau Public. I used the pytrends library to hit the Google Trends API, pulling hourly data for each city across all 8 emotional keyword categories. The trickiest part was normalization, Google Trends scores are relative, not absolute, so I had to pull each category independently to stop high-volume terms from drowning everything else out.

A few things worth knowing about the data:

✅ The Takeaway

At 3am, we're not who we pretend to be during the day. We're not productive. We're not scrolling mindlessly. We're scared. We're Googling whether our heart rate is normal. We're asking the internet questions we're too embarrassed to ask anyone else.

Data doesn't just tell you what people do. If you look closely enough, it tells you how people feel.

And apparently, at 3am, a lot of us feel like something might be wrong.