I Set Up Product Analytics in 3 Days With Google Sheets and Mixpanel's Free Tier. Here's the System.

When I joined Sonic Linker's founding team, we had a problem most early teams have: we were shipping fast, but we had no idea what was actually working.

We had no data engineer. No analytics budget beyond free tiers. And honestly, no time to wait for someone to build us a "proper" data stack.

So I built something scrappy. It took three days, cost us nothing, and we used it every single morning for the next six months. Here's exactly how I did it.

Start with Five Events, Not Fifty

The biggest mistake I see teams make is trying to track everything from day one. They instrument 40 events, half of them fire incorrectly, and two weeks later nobody trusts the data.

I started with five core events that mapped directly to our activation funnel:

1. Account created (obvious, but you'd be surprised)
2. First link shortened (our core action)
3. Dashboard opened (were people coming back?)
4. Link clicked (did the product actually work?)
5. Workspace invited user (our best retention signal)

That's it. I used Mixpanel's free tier (up to 100K events/month, more than enough for early stage). I worked with our dev to instrument these five events in one afternoon. We used their SDK, dropped the snippet in, and called it done.

The key was that every single one of these events answered a specific question I had about user behavior. Not hypothetical questions. Real ones I was asking in standups.

Build a Daily Dashboard in Google Sheets

Mixpanel has dashboards, but I wanted something I could customize without fighting their UI. So I set up a Google Sheet that pulled data via their export API every morning at 8 AM.

I tracked:

• DAU/WAU/MAU (basic health)
• Day 1, Day 7, Day 30 retention cohorts
• Time to first link shortened (activation speed)
• Links created per active user (engagement depth)

This wasn't fancy. It was literally five charts on one tab. But I looked at it every single day, and so did the founders. That's the point.

The Sheet also had a second tab where I manually logged qualitative notes from user calls. Sounds dumb, but pairing "retention dropped 8% this week" with "three users mentioned the onboarding was confusing" made patterns obvious way faster than staring at charts alone.

Use Segment's Free Tier as a Safety Net

Here's the thing: Mixpanel is great, but if you hardcode everything directly into their SDK, you're locked in. If you want to add Amplitude or PostHog later, you're re-instrumenting everything.

I used Segment's free tier (up to 1,000 users) as a passthrough. Our devs sent events to Segment, and Segment forwarded them to Mixpanel. It added maybe 30 minutes of setup, but it meant we could swap tools later without touching code.

At Finvestfx, we did exactly this. We started with Mixpanel, then added Amplitude when our data needs got more complex. Took me 20 minutes to flip the switch in Segment. Zero dev time.

One Person Owns the Data (And It Should Be You)

This is non-negotiable. Early on, someone has to own making sure the events are firing correctly, the dashboard is accurate, and the data actually gets used in decisions.

At Sonic Linker, that was me. Every Monday, I did a 15-minute data hygiene check:

• Spot-check event volumes (did "link shortened" spike or drop weirdly?)
• Test one user flow myself and verify events fired
• Compare our Sheet dashboard to Mixpanel's raw numbers (did the API pull work?)

This sounds tedious, but it's the only way to catch issues early. We once shipped a feature and didn't realize the tracking was broken for three days. Lost that data forever. Never made that mistake again.

Know When to Graduate

This system worked for us until we hit about 2,000 active users. At that point, Mixpanel's free tier wasn't enough, our event volume was growing, and we started needing more complex analysis (funnels, cohort breakdowns, multi-touch attribution).

That's when you graduate. Not before.

At Finvestfx, we eventually moved to a proper warehouse (BigQuery) and hired a data analyst. But we didn't do that at 500 users. We did it at 5,000, when the scrappy system was actively holding us back.

What Actually Matters

You don't need perfect infrastructure. You need a system that answers the three questions every early PM is asking:

1. Are people using the product?
2. Are they coming back?
3. What's breaking?

Five events, a Google Sheet, and 15 minutes a week will get you 80% of the way there. The other 20% can wait until you have money and headcount.

I've set this up twice now. It works. And honestly, I trust this more than half the over-engineered analytics stacks I've seen at later-stage companies.