The Common Metrics Every Product Runs On

By the end of this lesson, you should be able to: name the common metrics that tell you how a product is doing, pick a North Star that fits the type of product, and walk any product from acquisition to revenue with the right metric at each step.

This lesson is about common metrics in data science interviews. And honestly, I think this is the most important topic in the whole interview process.

Here's why. Any product case interview, or really any interview that touches product, is going to ask you how you would measure something. How do you measure whether a product is doing well? How do you measure whether the thing you built actually worked? There's a famous line in management: if you can't measure it, you can't improve it. And that's exactly what the interviewer is testing.

But it's not just interviews. On the job, this is the most important part of the work too. When you present to leadership, they don't want to hear how complex your analysis was. They want to know what moved. Getting visibility, showing your business impact, all of it comes down to your metrics. As a great data scientist, you need to know how to pick the right ones.

A lot of people struggle here. They know a few metric names, but when it comes to picking the right metric, defining it properly, and turning it into a clear answer, they get stuck.

So over the next few lessons, we're going to go deep on metrics. In the next one, I'll walk you through how to define a metric properly.

This lesson is the overview. It's a general metrics framework that works for any product. I'll show you the most common metrics, what each one tells you, and the traps worth knowing about. Let's get started!

The overall health of a product

The first metrics almost everyone looks at are active users. Daily active users, or DAU, is how many unique people used your product on a given day. Weekly and monthly active users, WAU and MAU, are the same idea over 7 days and 30 days.

These sound simple, but they're not. What counts as active? Is it just opening the app, or does the user have to actually do something, like watch a video or send a message? And whose day is it? Is it cut in UTC, or in the user's local time? Different companies answer this differently. We'll get deeper into that in the next lesson.

Next is the stickiness ratio. That's DAU divided by MAU.

Say your product has 2 million daily users and 10 million monthly users. That's 20%. In plain words, the average monthly user shows up about 6 days a month. For a real example, Duolingo reported about 37 million daily users and 113 million monthly users in the third quarter of 2024.1 That's a stickiness of about one in three.

So stickiness tells you whether your product is a daily habit, or something people only open now and then. A messaging app should be used every day. A travel app, maybe a few times a year. So it makes sense to only compare it against products of the same kind.

The North Star depends on the type of product

On top of these, most companies pick one North Star metric. That's the single number that best captures the value users get, and that drives the business long term. And here's the key point. The North Star depends on the type of product.

For a two-sided marketplace like Uber or DoorDash, it's the total value of everything ordered on the platform, usually called GMV. DoorDash calls it gross order value, and Uber calls it gross bookings.23 Both also track how many orders or trips get completed, and that only grows when there's enough supply and demand on both sides.

For a social or content platform that makes money from ads, it's how many people show up every day, and how much ad revenue each of them brings in. Meta reports daily active people across its apps and average revenue per person. Reddit reports daily active uniques and revenue per user.45 Dig one level deeper into ads, and Meta also reports ad impressions and the average price per ad.

For a SaaS product, it's net revenue retention. That's how much revenue you keep and grow from the customers you already have. Snowflake and Figma both report it.67 For a consumer subscription like Spotify, it's paying subscribers and revenue per user.8

For e-commerce, it's the total value of everything sold, how many active buyers you have, and how often they come back. Etsy, for example, reports gross merchandise sales, active buyers, and what it calls habitual buyers, the people who buy again and again.9

And for a free mobile app or a game, it's daily active users, how much time they spend, and how much they pay per daily user. Roblox, for example, reports daily active users, hours engaged and bookings per daily active user.10 Day 1, day 7 and day 30 retention matter a lot too, especially for a new game. And the bar is lower than you'd think. GameAnalytics looked at more than 16,000 mobile games in 2025, and the median game kept fewer than 4% of its players by day 7.11

Notice the pattern. Every one of these counts value users actually got, or money that actually came in. Not what you can simply buy. You can buy installs. You can buy sign ups. You can't buy someone coming back next week because the product was good.

Acquisition

Acquisition metrics tell you how many new users are coming in, and from where. Visits, sign ups, app installs, broken down by channel like paid ads, search, referrals, organic. Then there's customer acquisition cost, or CAC. That's how much you spend to get one new user.

Now, a few traps here.

First, a cheap channel isn't a good channel if the users it brings in never come back. Quality matters more than volume. So don't stop at counting new users. Check how many are still around a few weeks later, channel by channel.

Second, be careful when a channel looks better just because its users are older. Users who joined a year ago have had a year to spend money. Users who joined last week haven't. So always compare channels across users at the same age, like each channel user's first 30 days.

Third, a sudden jump in new users isn't always growth. It could be a promo, a one-time partnership, or a change to a default setting. Ask whether those users chose your product, and whether they stay.