Ninety Days Blind: The Season I Ran My Website Without a Single Number

On the first Sunday of March, I did something that felt mildly illegal: I deleted Google Analytics, unplugged the heatmap script, and cancelled my uptime monitor's weekly email digest. Then I sat there for a solid minute, staring at a website that had suddenly gone silent.

No dashboard. No bar charts. No little green line telling me whether anyone, anywhere, had loaded my about page since breakfast.

The rules were simple. For ninety days, I would run albertusvembrianto.com with zero data. If I wanted to know something, I had to either ask a human or guess. Any redesign, any new essay, any tweak to the homepage — all of it would be driven by taste, instinct, and the occasional email from a stranger.

Ninety days later, I'm ready to file the report.

The First Two Weeks Were Withdrawal

I won't pretend it was smooth. For the first several days, I caught myself opening my analytics dashboard out of pure muscle memory, the way you reach for your phone when a commercial comes on. The page would load, show nothing, and I'd feel a strange little pang — like walking into a room where furniture used to be.

What surprised me was what that pang revealed. I wasn't checking analytics because I had a question. I was checking because checking felt like work. It was the writer's equivalent of straightening pencils. A numbers ritual that produced the sensation of progress without requiring any actual decisions.

Once the ritual was gone, I had to face an uncomfortable question: what do I actually do on this website if I'm not managing it?

The answer, it turned out, was write on it.

Decisions I Made With My Gut Instead of a Graph

With no data to consult, every choice came down to one question: do I like this?

Some examples from the experiment:

• I rewrote my homepage headline because the old one read like it had been focus-grouped by a software company. The new one sounds like me.
• I killed a "featured projects" carousel that I'd always found clunky but had defended because "engagement metrics looked fine." Turns out I just didn't like it. It's gone.
• I published an essay I would have normally agonized over, because there was no dashboard to tell me whether the last three essays had "performed." I just wrote the next one.
• I moved my contact link from the footer to the top of every page, on the theory that if I were visiting someone's site and wanted to say hi, I'd resent having to scroll.

None of these decisions would have survived a meeting with my old metrics. The carousel had numbers on its side. The headline had a lower bounce rate than the replacement, probably. But "the numbers were fine" is a terrible reason to keep something you find ugly on your own personal website.

The Metric That Replaced All the Metrics

Here's the thing nobody tells you about deleting analytics: people will still tell you what they think. You just have to make it easy for them.

Over the ninety days, I received fourteen emails from visitors. That's it. Fourteen. In the same period last year, my dashboard claimed something like eight thousand sessions.

And yet those fourteen emails taught me more than eight thousand sessions ever did. One reader told me my about page made her laugh out loud on the L train. Another asked why I hadn't written anything about a project I'd mentioned in passing — so I wrote it, and it became one of my favorite pieces. A recruiter said my site "felt like a person," which is the single best performance review this website has ever received.

Fourteen data points, each one rich enough to act on. Eight thousand sessions, each one a shrug.

What Actually Got Worse (Because I'm Being Honest)

This wasn't all a glowing tale of liberation. A few things genuinely declined:

I published less consistently at first. Without the weekly traffic report as a pseudo-deadline, some weeks slipped. The numbers had been functioning as a guilt mechanism, and guilt, it turns out, is a real (if ugly) motivator.

I made at least one objectively bad call. I reorganized my navigation based purely on vibes, and a friend later told me she couldn't find my writing for a full month. A heatmap would have caught that. I fixed it, but slower.

I lost all sense of scale. I have no idea whether the site had a good quarter or a terrible one. For a personal site, I've decided that's acceptable. For anything with stakes, it would be malpractice.

What I'm Keeping Now That the Experiment Is Over

I reinstalled a privacy-respecting analytics tool when the ninety days ended, mostly out of curiosity. I checked it once. Traffic was roughly the same as always — the internet barely noticed my experiment, which is exactly as it should be.

But the dashboards don't run the place anymore. I check the numbers about once a month, the way you'd glance at a thermostat, not the way you'd stare at a stock ticker. The fourteen-emails standard stays. If a piece of writing doesn't move at least one person enough to write to me, no chart is going to convince me it mattered.

Here's my takeaway, for whatever it's worth: analytics tools were designed for sites with millions of visitors, where individual humans are statistically invisible and aggregates are the only honest picture. A personal website is the opposite situation. You have dozens of visitors, maybe hundreds. Every single one of them is a person who chose to show up.

You don't need a dashboard to see them. You need a site worth writing to, and an inbox that's actually open.