How to Choose the Right Product KPIs

What are product KPIs? Why do they matter?

Product KPIs are the numbers you look at every day that tell you how well your product is achieving its goals. They help you judge whether your strategy is improving the product over time, keep your efforts focused on what matters to the business, and evaluate the outcome of experiments. Good stories built around KPIs also help you make the case to company executives — about the additional resources your team needs, about why you’re not prioritizing their pet feature, or about why you deserve a raise and promotion.

KPIs will only be useful if they are specific, measurable, and actionable. They need to be clearly defined: no one should be arguing about, for instance, what counts as a daily active user of your service. They need to be easy to track: ideally, they should flow directly from instrumentation in your product rather than requiring labor-intensive data gathering or aggregation. And they need to be (at least partly) driven by product decisions — only then can inform your choices.

Note: Product KPIs are not the only numbers that matter. You’ll also want to monitor KPIs for particular features and launches and business metrics like customer lifetime value and acquisition cost (which are important but require a lot of assumptions to calculate, and therefore differ from the directly measurable KPIs we’re discussing here).

How many KPIs?

How many KPIs should you monitor? As few as you can while still getting a clear picture of how your product is doing. For a complicated product, you may well have a dashboard with 6-8 metrics that you look at regularly and investigate any unexpected changes in.

That said, if you have more than 2-3 KPIs, you also need prioritization within your list. Most things you change about the product will make some KPIs better and others worse. It’s easier to accept these trade-offs if you’ve already thought about what matters most.

We'll talk more about when data lies later, but even your most important KPIs aren't foolproof. A robot can make purely data-driven decisions; your value as a PM comes from combining data with your qualitative understanding of what your users need. Data tells you what people do; empathy and user research tell you why.

Rule #1: Impact to the business, the customer, or both

You only get a handful of KPIs, so you have to choose ones that count. There are only two reasons to include a KPI on this shortlist: (1) it’s moving the needle for your company, or (2) it’s moving the needle for your customers.

KPIs that move the needle for the company are tied to how you make money. Let’s say you’re the PM for an ad-supported consumer app. Your top business concerns are likely how many users your product has and ad revenue per user. Because ad revenue per user may be driven more by ad sales and pricing than your product, you might track time spent in the app or number of ad-containing screens visited per user per day instead. If your product isn't generating revenue yet, the usage or engagement numbers that help you raise funds should be your business priority.

KPIs that move the needle for your customers ensure you are delivering value — this is what enables you to make money long-term, not just short-term. Consider the same app as above. You have two sets of customers: consumers and advertisers. You might track consumer engagement via retention rates and advertiser value delivered through ad click-through or conversion rates, or even (if your ads platform is sophisticated enough) return on ad spend.

Note: In this example, advertisers make money for your company, and ad revenue numbers may be one of your KPIs. But it's important to also keep track of how well your product is serving advertisers — which is not the same as how much money you're getting from them.

Rule #2: Know which way is good

A good KPI must be unambiguous -- you should know which way you want it to move. This seems obvious, but breaking this rule is a common pitfall.

For instance, “total time spent on the app” is not a great KPI for a productivity-oriented product like an email or to-do list application. What if product improvements reduce the number of clicks required for users to accomplish key tasks, or automate processes users previously had to complete manually? These features may improve the appeal of the application and increase user acquisition and retention, but they will also decrease the amount of time each user spends in the app. The total time spent could go up or down. If you want these effects, you should instead track the number of users and average time spent as two separate KPIs — looking for an upward trend in users and a downward trend in time spent.

Another example is a system to improve street parking for drivers and parking enforcement efficiency for traffic police. You might be tempted to measure effectiveness in the form of “number of tickets issued.” But improving the parking situation for drivers could lead to fewer people parking illegally (moving this number down) while improving enforcement could help us catch a larger proportion of offenders (moving this number up). It would be better to track tickets issued (or ticket revenue, depending on your goals) per dollar spent on enforcement.

Rule #3: Responsive, but not volatile

You’ll check your KPIs daily, and you’ll use them to evaluate A/B test results, advocate for your product roadmap, and set alerts. Others in the business — including your boss and higher-up execs — may watch their daily moves as well. You want your KPIs to respond to product improvements: if you make the product better, your KPIs should reflect that. However, you don’t want to field phone calls or deal with midnight alerts because of natural variability in your KPIs. For instance, since email applications see many more daily active users (DAUs) on weekdays than weekends, a seven-day trailing average of DAUs is a more appropriate KPI than raw DAUs.

Rule #4: Better to lead than lag

There are three types of indicators: leading, concurrent, and lagging. Leading indicators predict success or failure; concurrent indicators tell you that you're succeeding or failing right now; lagging indicators tell you you've already succeeded or failed. Since leading indicators provide feedback sooner in your work cycle, they can save you time and effort.

By definition, most KPIs instrumented in your product will be lagging indicators of the quality of your product roadmap. (You'll only see changes in KPIs over time after you've made decisions and launched features.) However, if you have specific goals, you may be able to pick leading indicators of achieving those goals. Let's say your goal for the year is to reach one million DAUs. DAUs themselves are a concurrent indicator of this goal. But customer acquisition and churn numbers could serve as leading indicators. Your choices will move these KPIs more quickly than overall DAUs, and these numbers enable you to predict whether you're on track to hit your goal.

Real-world KPIs

A consumer email app with 20M+ users

• 7-day rolling average DAUs
• 7-day and 28-day retention rates
• Google Play Store ratings (found to be leading indicator of DAUs)
• % of users composing emails (found to be predictive of long-term retention)

An online advertising platform

• Monthly run rate compared to the same month prior year (to account for seasonality)
• Advertiser cost per click (CPC) and cost per acquisition (CPA)
• Advertiser retention, upsell, downsell, and churn (same advertiser versus previous month)
• Advertiser satisfaction scores and net promoter scores (surveyed quarterly)

An internal infrastructure tool

• Daily number of users and number of teams
• 7-day and 28-day retention rates (internal)
• Referral rates (internal)
• Ratings of product (solicited via in-product prompt)
• Monthly internal survey of value added / time saved to other teams

An email campaign for an e-commerce website

• Email open rate
• Click-through rate (CTR) on links
• Conversion rate: how many users purchase an item after receiving the email
• Unsubscribe rate

A SaaS company with a freemium model

• Revenue run rate
• Monthly active users for free version of product
• Conversion rate from free version to paid version
• How many days (of last 7 and last 30) user was active on platform

Common KPIs

Pick the right KPIs for your product, not a “standard” set of KPIs for your solution. (Don’t measure “app KPIs”; measure how well your app is doing its job!) That said, here’s a bank of common KPIs to kick-start your brainstorming.

Consumer-focused:

• Daily Active Users (DAUs), current or rolling averages
• Monthly Active Users (MAUs)
• Retention rates (usually 3 days, 7 days, 14 days, or 28 days after joining)
• Average time spent on site/app per user
• Sessions per user (per day or week)
• Engagement metrics (e.g., “likes” or emails composed) per user

Advertiser-focused:

• Click-through rates (CTRs)
• Conversion rates (CVRs)
• Cost per sale or install (CPA/CPI)
• Return on ad spend (RoAS)