The behavioral analytics guidePublished date: Aug 1, 2018
Ask the top organizations around the world—in any industry—how they build successful products, and you’ll hear some variation of this answer: “We followed the data.”
When we study user behavior, we gather data on what people do—and what they don’t do—so we can build products that people will value. After all, you wouldn’t build a product that no one would use, nor would you invest time, money, and energy in a feature set that has no observable demand.
Behavioral analytics helps you understand what your users and customers are doing and why, so that you can build better products.
What is behavioral analytics?
Behavioral analytics is the process of using behavioral data—or data generated by, or in response to, a customer’s engagement with a business, such as page views, email sign-ups, or other important user actions—to construct behavioral cohorts and analyze things like engagement, conversion, and retention.
When behavioral analytics is applied to digital product management (often called product analytics) it can take the guesswork out of development and mitigate the risk of every bet you make. Teams use behavioral analytics to determine what users like and don’t like and, by inference, what adjustments will make the product more valuable.
Behavioral analytics is an important part of building best-in-breed products.
Why is behavioral analytics different than other types of business analytics?
What sets behavioral analytics apart from other types of business analytics? It combines two technologies: user segmentation and behavioral or event tracking.
While some analytics vendors offer only one or the other—user (object) data or behavioral (event) data—behavioral analytics ties users to the events they trigger, thus producing a map of their actions as a flow or customer journey.
Viewed either alone or in aggregate, user journeys tell stories that teams can use to tweak and improve their product development, marketing, and launch strategies.
The importance of user properties and event properties
Behavioral data can, and should, always be tied back to a single end-user, but good behavioral analytics typically involves separating “user properties” from “event properties”. Both are very important to analyze because they tell you who your users are and what they’re doing.
User properties are anything around user traits that reflect the state of the user and apply across all their events, including:
- Unique vs non-unique
- Active vs inactive
- New or not new
There are also a number of custom properties you can analyze, depending on the type of app or product you have. In a gaming app, for example, your custom user properties may include an amount of in-game currency and current game level.
Event properties involve any action a user takes in your product, which could be anything from making a payment to finishing a level to clicking a button. Event properties show you how your users are engaging with your app and will, of course, vary depending on the type of app you have and the specific information you think is required for understanding a particular event.
How can you benefit from behavioral analytics?
Whether you sell socks, streaming video, or a B2B SaaS product—if you want to grow, you need to do (at least) two things really well:
- Acquire new users
- Retain those users
As Streaming, Sharing, Stealing co-author Michael D. Smith explained to The Signal, if a company wants to personalize its service to users—thus increasing the likelihood of acquisition and retention—it needs users’ behavior data. For example, a streaming movie platform can’t know that a user loves horror films simply based on their age, gender, or nationality. The key to that kind of understanding is seeing the signals a user sends with their behavior.
Behavioral analytics can provide user-level data so teams can answer such questions as:
- What do users click within the product?
- Where do users get stuck?
- How do users react to feature changes?
- How long do users take from first click to conversion?
- How do users react to marketing messages?
- Which ads are the most effective?
- How can we nudge users to be more successful in what they’re trying to accomplish?
Conducting behavioral analysis is more complicated than simply running reports in an analytics tool. Teams must first identify what they want to achieve and write down the paths they expect users to take. Only with preset expectations can teams identify whether users are deviating from the ideal path, and then redirect them.
The tech-driven insurance provider Lemonade adjusted its user paths to increase revenue. The goal was to convert more website and app visitors into paying customers. With Mixpanel analytics, they noticed a “staggering drop-off” in user flow right before the point of purchase. By analyzing the page where the drop-off occurred, the team realized it was due to a technical error and a weak call-to-action (CTA). They fixed the bug and reworded the CTA, which led to a 50 percent increase in the number of users who purchased additional coverage.
Steps for successful behavioral analysis
Embedding behavioral analytics into your organization’s culture can seem daunting—it involves a lot of decisions (big and small), followed by many tweaks and optimizations. As with most things, a “one-size-fits-all” approach isn’t going to work.
Some of the work involved is technical (wrangling data, implementing a tracker), while other tasks are strategic (goal setting, building a tracking plan) and still others are analytical (building and interpreting reports). There’s plenty of overlap, and achieving success requires tight collaboration between teams and departments.
Whether you’re starting from zero or already have a fully-fledged behavioral analytics stack in place, here are the high-level steps for kicking off any behavioral analytics project:
1. Select goals, KPIs, and metrics
To determine whether users are reaching the right goals (e.g., purchase or conversion), select the KPIs and metrics that indicate progress toward those goals.
- A fitness app that makes money through monthly subscriptions tracks paid subscriber growth.
- An enterprise resource planning (ERP) software that relies on annual contracts tracks users that complete the onboarding sequence.
Note that behavioral analytics goals might be a bit different than you’re used to; we tend to conflate goals and results, and it’s much better to start from a goal centered on measurement rather than a goal centered on a particular result.
For example: “Increase sign-up conversion from 5% to 10%” is a wonderful goal in general—but it’s not an analytics goal. It’s a result.
Why? Because measuring something is not the same as changing it. The macro goal of any behavioral analytics project is to accurately capture behaviors, so we can analyze those behaviors and then make informed decisions about what we may change to make an impact.
A revised version of this goal might be: “track all user journeys from Landing to Sign Up and find the best/worst conversion paths.” The revised version is a better goal, because it focuses on measurement rather than a result.
2. Map out the critical paths
Critical paths are the series of events that need to occur or actions your users need to take for your product to be successful.
To determine your critical paths, ask, “What are the most common paths for users to reach their goals based on this service or app’s design?” If the product has already been launched, you can use actual user data to answer this question. If the product is pre-launch, use wireframes of the suspected or intended flow.
Critical path examples:
- An e-commerce website tracks a user from their first page visit to adding an item to their shopping cart to checkout because that flow leads to purchases.
- A streaming music app can track users as they move from its homepage to playing a song and, hopefully, purchasing that song.
3. Create a tracking plan
Based on the user flow, decide which events you need to track. Tracking everything is a mistake—too much data clutters the analytics and makes useful information harder to find.
Some events contain multiple properties. For example, the event for playing a song within a music app could contain properties for the song title, genre, and artist.
To keep events and properties organized, companies typically create a tracking plan in a spreadsheet. This directory of all events serves as a map for implementing the analytics tool (see Step 5, below). A tracking plan should be revised and updated as the product, team, and goals change.
(To reduce the burden of trying to share and control access to the spreadsheet, Mixpanel’s Lexicon feature stores the event name taxonomy for all to see.)
Involve all teams—analytics, product, marketing, and engineering—in drafting the tracking plan. They’ll all need to understand how the users and events are named and organized if they’re going to run reports and understand the results.
4. Use unique identifiers to understand behavior across platforms
Most digital products today exist across multiple platforms, which makes it difficult to track unique users and understand their behavior across these platforms. One user can appear to be multiple people unless assigned a unique identifier—either an email or string of characters—that persists across platforms and devices and connects the touch points along their journey.
Teams should ensure their behavioral analytics platform vendor provides a unique identifier that won’t change over time so that they can track and understand their behavior across platforms.
5. Implement analytics and begin event tracking
Once the tracking plan is complete, you can deploy behavioral data analytics software and use its SDK or API to integrate it with your products. Assign a unique identifier for users and set up user and event properties as outlined in the tracking plan.
Sometimes you’ll discover additional events you want to track during implementation. This isn’t an issue as long as you update both the tracking plan and the analytics service.
When all your tests pass, you’re ready to start collecting real data and analyzing users.
Behavioral data and product management
Behavioral data is primarily used for product management. The ability to apply the results of behavioral analytics—using data to build better products while mitigating the risk of trying something new—is the return on your investment.
To apply the results of behavioral analytics, start with a hypothesis, use several different metrics to validate it, and then (through deduction and exploration) construct a KPI based on what you’ve learned. Only then are you ready to experiment with new features, test them on key cohorts of customers, and finally declare (with confidence) that you are building a better product.
This process is designed to be repeated. And as you repeat this cycle, you successfully iterate on your product’s development with a statistically-driven guarantee that the investments you’ve made are the right ones.
Most teams use segmentation: joining users characteristics and behaviors. An e-commerce app, for example, could create a segment for recent users who abandoned items in a shopping cart, or filter for power shoppers who access the app multiple times a week.
Segmentation helps teams build more complete customer profiles and create cohorts (defined segments of users based on common important characteristics or experiences). Armed with this information, they can adjust their product and marketing to better address the specific needs of each segment or of the business.
STARZ PLAY segmented users that signed up through its free trial offer and found that some users were gaming the system for multiple trials. By creating alerts for the negative behavior, the product team closed the loophole and saved 8x on its marketing spend.
To track users’ progress toward outcomes such as purchases or signups, many teams use funnel reports. Funnels display a series of stages in a user journey, as well as how many users progress from one stage to the next—for example, from download to signup and purchase. If one stage has a low conversion rate, it’s a signal that that stage needs attention.
Funnel data lets teams A/B test different buttons, messages, and images, to see if small changes improve conversions. Funnel data can also be used to personalize parts of an app or website, say, to greet returning users, or present prospects with content relevant to their industry.
Teams can also use funnel data to deduce which behaviors are correlated with high retention. Your retention rate shows you an average across all users. By digging into behavioral and demographic trends beneath the metric, or by looking at specific cohorts, you can identify what makes your users stick around, and which users are likely to become loyal customers.
A media site, for instance, could look at the cohort of users that continue to return to the site eight weeks after signup to see if they share certain behaviors, such as a propensity to leave comments.
With a clear view of what’s happening within the product, teams can run experiments and make alterations to improve the product and help users find more value.
This is what “following the data” really looks like.
How to choose the right behavioral analytics provider
The best behavioral data analytics solutions are versatile and compatible with many systems, and offer quick results. They let teams integrate their full range of digital products, from desktop and mobile applications to internal services. They have friendly, intuitive interfaces that allow teams to quickly find answers, and they provide a breadth of functionality, including some that teams don’t yet need, but may in the future.
Swapping out a behavioral analysis platform can be a costly and time-consuming endeavor. Teams can save themselves time and money by doing their research and investing in a platform that offers them room to grow into.
Look for user behavior analytics tools with:
- Ability to automatically capture user and event data points
- Ability to access data in real-time and query it in a variety of ways
- Pre-built reports such as funnels, cohorts, and retention
- Guardrails, automated notifications, and recommendations
- A data visualization component
Take the next step
Maybe you’ve been hearing things like “we don’t have the data” or “we don’t have the resources.” If you read between the lines, what you’re really hearing is: “Behavioral analytics is not a priority.”
This is a very risky attitude; if you don’t have the data, you can’t follow it, and therefore every decision your company makes basically amounts to crossing your fingers and hoping it works.
The best way to make behavioral analytics a priority is to start talking about it. If it helps, our customers have lots to say. And if you have questions, don’t hesitate to get in touch with an expert.