A Product Metrics Primer: How to Choose, Refine, and Track the Right Metrics

Based on Shreyas Doshi's framework — originally shared as a Twitter thread in September 2020.

Metrics are central to modern product management. Yet most product teams either track too many metrics (drowning in dashboards), or too few (flying blind). Shreyas Doshi — former PM leader at Stripe, Twitter, Google, and Yahoo — has spent over a decade refining how he thinks about product metrics. What follows is his primer, distilled into a practical guide.

The Six Product Metrics Categories

When conceiving your metrics, consider each of these categories and pick the right metrics across them. They cover multiple granularities and perspectives you need for rigorous product decisions.

1. Health Metrics

Key question: Is the product available and performing in the manner that users would reasonably expect?

Health metrics are foundational. If your product is slow, unreliable, or losing data, nothing else matters. These are the vital signs.

Examples: latency, initial load time, uptime, data loss rate, HTTP errors, crash rate.

2. Usage Metrics

Key question: How are users actually using the product?

Usage metrics reveal behavioral patterns — they tell you what users do, when they do it, and where they get stuck. They're your window into the reality of how your product lives in the wild.

Examples: time-of-day and day-of-week usage trends, top N actions, funnel metrics, help docs usage, retention curves, password reset rate.

3. Adoption Metrics

Key question: Is the product (along with its key features) being used as much as we'd hope and in the ways that we'd like?

Adoption metrics sit one level above usage. They tell you whether new users are sticking around and whether strategic features are gaining traction — not just whether people are clicking around.

Examples: active users, DAU:MAU ratio, N-of-M-day usage, strategic feature adoption trends, free-to-paid conversions.

4. Satisfaction Metrics

Key question: What is our customers' overall sentiment towards the product or its main features?

Quantitative data tells you what is happening; satisfaction metrics start to tell you how users feel about it. These metrics are a bridge between behavioral data and qualitative insight.

Examples: overall CSAT, new feature CSAT, support CSAT, NPS, user-reported satisfaction scores.

5. Ecosystem Metrics

Key question: What is the macro state of the product within the domain in which it operates?

Ecosystem metrics zoom out from your product to its position in the broader market. They can be either leading or lagging indicators for your core business metrics.

Examples: share of wallet, third-party integrations, industry rank, market share within target segments, percentage of TAM captured, brand awareness.

6. Outcome Metrics

Key question: What overall results are we seeing from this product?

Outcome metrics are the bottom line. In many cases these are about revenue and profit, but the right outcome metric depends on your product's mission — for a medical product, it might be "number of lives saved."

Examples: revenue, margin, revenue per user, active users, market share, transactions, percentage of Fortune 1000 covered.

A Note on Overlap

A given metric can belong to more than one category. That's fine. You may also skip a category altogether if it doesn't apply to your product. The categories are a thinking tool, not a rigid taxonomy.

Choosing Your Key Metrics

From the universe of possible metrics across these six categories, you need to select a small set of Key Metrics (KMs) — typically 3 to 5. These are the numbers your team rallies around, the ones your execs and board will look at to assess the product's progress.

A good Key Metric meets four criteria:

Responsive to product changes — it moves when you ship meaningful work.
An aggregate measure of value — it reflects the product's overall value to its users, not just one narrow slice.
Readily tied to business value — the connection to the company's goals is clear.
Expected to be long-lasting — it should remain relevant for at least two to three years, not just the current quarter.

Key Metrics give you a great high-angle view of how the product is doing. You'll typically set annual goals and targets around them.

The Case for Leading Metrics

In certain cases — B2B SaaS products with long adoption cycles being a prime example — Key Metrics can be lagging indicators of work done. Revenue might take months to reflect a product improvement. That's why you also need a small set of Leading Metrics (LMs): indicators that move faster and give you earlier signal on whether you're headed in the right direction.

Define 3 to 5 Leading Metrics in addition to your Key Metrics if this applies to your product. Leading Metrics might change more frequently than Key Metrics, and that's expected — they're meant to be more dynamic.

Across both B2B and consumer products, 3–5 Key Metrics and 3–5 Leading Metrics have worked well in practice.

Metrics Ownership: Don't Over-Optimize for Control

Some product managers perform major contortions to avoid owning metrics they don't fully control. "I depend on Sales for this, so I can't use it as a Key Metric." Shreyas pushes back on this instinct. There's no such thing as 100% control over any metric. If Sales can't or doesn't sell your product, that's still your problem. Own the metrics that matter, even when they depend on cross-functional partners.

Practical Tactics for Staying Metrics-Aware

Having the right metrics is only useful if you actually look at them. A few bonus tactics:

Treat your metrics dashboard like a product. Invest in making it clear, useful, and up to date.
Track your dashboard's usage. Yes, it's meta — but if nobody's looking at the dashboard, that's a signal.
Set a Chrome pinned tab for your dashboard. Make it impossible to forget.
Review it regularly. Every morning, or at minimum every few days.
Set a weekly email report. A push mechanism ensures the metrics come to you even when you're busy.

The Most Important Principle: Be Metrics-Informed, Not Metrics-Driven

This is the thread's parting — and arguably most important — idea.

Metrics are one of the inputs you need to make product decisions and assess progress toward your company's mission and strategy. They are not the only input. Also use qualitative research, customer conversations, strategic intuition, and competitive context.

If your team doesn't understand the "why" behind a given metric and yet proceeds to impact that metric, the team is likely being driven by metrics rather than being informed by them. That's a dangerous place to be.

The ability to know when to break the rules — and to actually break them — is often what distinguishes great PMs from the rest. A great PM will diligently track metrics but will rely on them as one source of input among many.

One more thing worth remembering: "Metrics made me do it" is not an acceptable answer from a leader when something goes wrong. In those situations, you probably don't need different metrics — you need a different leader.

Shreyas Doshi is a product advisor and former PM leader at Stripe (where he was the first PM manager and helped grow the PM team from ~5 to 50+), Twitter, Google, and Yahoo. You can follow his writing at @shreyas on X.

Originally published on X.