The complete guide to product analytics

Alright, so you understand why product analytics is important. You want to know how to actually do it. We'll walk through five steps of increasing sophistication to get you up to speed.

Step 01 (Beginner): Tracking user behavior
Step 02 (Beginner): Defining key adoption metrics
Step 03 (Beginner): Monitoring key metrics
Step 04 (Intermediate): Segmenting users into personas
Step 05 (Advanced): Just-in-time experiences

Step 01 (Beginner): Tracking user behavior

Product events, or user actions, take place across various platforms, including websites, software products, and marketing channels like email. Tracking these events is the first step towards product analytics.

Despite being the first step, most companies find tracking events to be hard. There’s a decision paralysis around naming conventions and which events to track. Our recommendations? Start with our list of 10 fundamental events. It’s a list of basic events that most B2B SaaS companies should be tracking.

The next level is to track specific events that are relevant to your specific product. The easiest way to do this is to use a tool that utilizes auto-track capabilities. Flywheel, Pendo, and Heap are all examples. Auto-track comes from a small code snippet that automatically tracks every user interaction where installed. This is a fantastic way to start learning about your users immediately. The downside of this approach — your data will be less formulaic and harder to aggregate at scale. Most companies start with auto-track and move towards manual tracking over time.

Examples of auto-track events look like this:

• “Clicked button with text Sign Up”: Occurs when a user clicks on a button that has the text “Sign Up”. Straightforward.

• “Played video”: Occurs when a user plays a video. Also obvious.

• “Clicked p with text Flywheel EA • See how it works”: Occurs when someone clicks a specific element on our marketing website. Still descriptive, but if this button changes then the event will no longer fire in the same way. This highlights some of the limitations of auto-track.

If you’re a larger company or have access to engineering resources, you can opt to manually track events. Some products, like Flywheel and PostHog, support both paths. Manually tracking events requires an engineer to write code that triggers events when actions are taken. Each event must be given a name, ideally a descriptive one, and is set up individually.

The Segment B2B SaaS spec is a collections of recommended, manually tracked events. Here are some other examples:

• "Page View" event: Occurs when a user visits a specific web page on your site. This helps you understand which pages are generating the most traffic. Most tracking snippets fire these events without requiring any engineering effort.

• "Sign Up Button Clicked" event: Triggered when a user clicks on the sign-up button. This can help you identify the effectiveness of your call-to-action (CTA) and registration process.

• "User Invited" event: Takes place when a user invites a team member to use your software or service. This event can give you insight into your product's vitality.

If you decide to track events manually, use a consistent naming convention. It should be a clear and consistent naming structure for your events. This will make it easier for team members to analyze the data and understand the metrics.

Step 02 (Beginner): Defining key adoption metrics

Now that you’re tracking basic events, it’s time to choose which ones that are most important. This will help you cut through the noise and focus on the user actions that truly matter. For a deeper look, check out our article on key product metrics for a full list of 39 to choose from, as well as how to decide which are most relevant.

To start, though, you’ll want to establish clear goals and metrics for product adoption, such as activation rate, feature adoption, and time-to-value. Look into these user journeys and decide which events correlate with success. Regularly monitor these metrics to evaluate the success of your adoption efforts and inform future improvements.

For example, Facebook is famous for tracking the number of friend requests sent by new users within a given time frame. Similarly, Jira keeps a close eye on the number of new signups that create 3 or more tasks in a new project. These are both user actions that occur relatively long after the initial signup event — but signal higher activation.

Pro tip: Expect major drop-off in your funnel. We’ve talked with hugely successful PLG companies that lose more than 60% of signups before they even get into the product. The culprit? A required email click to validate they're a real user. The volume of users that hit your key milestones may be very low.

Step 03 (Intermediate): Monitoring key metrics

This is where we finally get into what most people view as product analytics. Simply tracking and storing events won’t be very consumable or useful at scale, so most companies opt to use software for visualizing and querying the raw event data.

There are a few different approaches to monitoring metrics:

1. Product analytics dashboards

2. Manual queries

3. Periodic presentations

Product analytics dashboards:

Most companies decide to have some level of product analytics dashboard set up. We have an extensive set of dashboard best practices since the topic is large. Without going too in-depth, it’s normal to have a singular dashboard that contains your most important metrics, used alongside supplementary dashboards or charts for less critical metrics. Tools like Flywheel will build dashboards for you, or tools like Amplitude and Mixpanel allow for designing niche and in-depth dashboards.

Manual queries

Just like it sounds, these are queries that you run yourself. If you know SQL or other querying languages, you can directly access the raw data from your database or data warehouse and find answers to questions as they come up. Generally, there is a set list of saved queries that are run whenever you want the most recent update on a metric.

Periodic presentations

It's a bit of a stretch to call this a solution, but nonetheless it's common for executives to request presentations from data analysts. These presentations are generally held at a set cadence and go over key metrics, recent changes, and notable trends. The benefit of presentations is the ability to ask follow up questions and dive deeper into interesting insights. The negative is that you’ll often just get updates on the same metrics every time. Having a readily consumable dashboard replaces the need for this type of update and allows executives to spend more time diving deeply into anomalies. 

Step 04 (Intermediate): Segmenting users into personas

Now that you’re tracking events and monitoring key metrics, it’s time to level up. Specifically, by tracking user activity based on the types of users they are. Segmentation is the key to optimizing your product and go-to-market strategy.

You’re likely familiar with the concept of an ICP, or ideal customer profile. Find software that helps you build out granular customer segments, then analyze the adoption patterns of those users. You’ll often find that the feature adoption changes from persona to persona, helping you build accompanying content or in-product help guides that can be tailored to that specific audience.

This level of segmentation can also help companies decide which users are important to include in feature analysis. For example, you may have a product that targets Product Managers. Once you've segmented by personas, you might find Content Marketers rarely adopt your product and are lowering your overall activation rate. Try filtering out that segment and re-analyzing your product activation metrics according to your core audience.

Most product analytics tools allow you to build user segments in some fashion. The real difference comes down to ease of creation. We've found that most companies don't take advantage of user segmentation features due to how confusing it is in legacy tooling.

Step 05 (Advanced): Just-in-time experiences

The last stage, and the most advanced, is delivering messaging in your product to users based on:

• The customer segment they belong to

• Previous engagement with your product

• Overall company engagement with your product

This ladders into ABM tactics. You’ll no longer send irrelevant messages to users if you can avoid sending information about features they’ve already used. Additionally, if a content marketer hasn’t used a specific feature, but they also probably shouldn’t, you can reduce message spam. 

These messages can be delivered in many formats. Tooltips, modals, pop ups, and messages through platforms such as Intercom and Drift are common examples.

Let’s go back to Atlassian. Jira is most often used by engineers and product managers. However, many marketers log into Jira track which features will be delivered shortly. This helps marketers build out content according to a product launch plan. Once you understand the behavior and rationale of these marketers, you can avoid messaging them about new backlog grooming features. Instead, Atlassian could inform these marketers about alerting functionality to help them stay on top of upcoming releases, automatically.