The Only Three Metrics That Matter in Your First 90 Days as a PM

When I joined Sonic Linker as part of the founding team, I did what every good PM does. I set up analytics, built dashboards, created tracking for 20+ events. Weekly active users, session length, feature adoption, funnel conversion, retention cohorts, the works.

Two months in, I realized I was looking at numbers but making zero decisions.

The problem wasn't the data. It was that I didn't know the product well enough to know what mattered yet. I was drowning in metrics that all seemed important but none of them were actually driving what I did on Monday morning.

So here's what I learned: in your first 90 days, you don't need 20 metrics. You need three. And they're not the ones you think.

1. Time to First Value (However You Define It)

This isn't about activation rates or onboarding completion. It's simpler. How long does it take a new user to get something useful out of your product?

At Sonic Linker, we were building an AI tool for content teams. I spent the first month obsessing over signup-to-first-login rates. Meaningless. What actually mattered was signup to first AI-generated output they actually used.

Turns out, that number was 47 minutes on average. Way too long.

We cut it to 8 minutes by changing one thing: instead of asking users to connect integrations first, we let them generate something immediately with manual input. The integrations could wait.

That one metric, time to first value, told me where the product was broken. It forced conversations with users who dropped off. It made prioritization obvious. Everything else was noise.

The key is defining "value" correctly. Not a completed profile. Not a tour. The moment they got something they couldn't get before.

2. Repeat Usage Within 7 Days

Retention curves are great. Monthly cohorts are important. But in your first 90 days, you don't have months of data.

What you do have is this: did someone who used your product come back within a week?

At Finvestfx, we had 20+ enterprise clients using our forex platform. The actual transaction volume was steady, but I noticed something weird. Some clients would use the platform daily for trade bookings. Others would log in once, then disappear for two weeks, then come back.

I dug into the 7-day repeat usage. The daily users were small trading desks with high-frequency needs. The sporadic users were corporates doing monthly hedges. Same product, completely different use cases.

This changed our roadmap. We'd been building features for the high-frequency traders because they were loud. But the real revenue upside was in the monthly hedgers, we just needed to remind them we existed. We added a simple email digest with FX rate alerts. Repeat usage for that segment jumped 40%.

Seven days is short enough to spot patterns fast, long enough to separate real usage from tire-kickers. If someone doesn't come back in a week, they're probably not coming back at all.

3. One User Behavior That Predicts Success

This is the hardest one to find, but it's worth it.

At Sonic Linker, I noticed something after talking to our most active users. Every single one of them had saved at least three AI outputs within their first session. Not published, not shared, just saved.

Users who saved fewer than three? 80% churned within two weeks.

We didn't have fancy cohort analysis or predictive models. I just exported the data to a spreadsheet and looked for patterns. Saved outputs correlated with retention better than anything else we tracked.

So we changed onboarding to encourage saving. Added a "save for later" prompt after generation. Made the saved items library more prominent. That one behavior became our leading indicator.

The mistake most PMs make is looking for the *outcome* metric. MAUs, revenue, conversion rates. Those are lagging. By the time they move, it's too late to fix what's broken.

Find the behavior that predicts the outcome. That's your real metric.

Why Only Three?

Because in your first 90 days, you're still learning the product. You don't know what matters yet. If you track 20 things, you'll track nothing.

Three metrics force you to make choices. They force you to talk to users to figure out what "value" actually means. They force you to prioritize.

And honestly, three is about all you can keep in your head at once. If I can't recite my key metrics in a standup without looking at a screen, I'm tracking too much.

Once you hit 90 days, expand. Add your revenue metrics, your engagement funnels, your cohort retention. But in the beginning, stay focused.

The goal isn't to measure everything. It's to measure what tells you where to go next.