To stay competitive, it’s essential to understand how your customers are interacting with your brand across every touchpoint. Enter product analytics. Broadly, the goal of product analytics is to understand user behavior and identify which product experiences led to customer success and which led to customer friction (or even churn). For example, product analytics might be used to determine if users who completed a virtual tour of the product had a higher two-week retention rate than those who did not. Product analytics can help businesses answer questions like: What are users doing in the product? How do different segments behave and perform? When does attrition begin? And what are our funnels and pathways of engagement?
Not surprisingly, product analytics is heavily leveraged in departments charged with delivering products: product management, growth, design, and engineering. Marketing teams also leverage product analytics by treating a website or ad campaign as the product. And customer success teams can leverage product analytics for insights into customer usage to drive adoption and engagement, help customers unlock value from new features, and facilitate upgrades and renewals.
The current “first generation” of product analytics tools can help businesses improve the user experience and increase revenue. However, as the world moves increasingly online, customers expect more — faster response times, more personalized experiences, and a deep understanding of what they value. Businesses want to know not just what their users are doing, but why, so they can take action in the moment. Product analytics tools weren’t designed to address these needs because they only provide top-level visibility into the product. As a result, there’s an emerging next generation of product analytics: a new breed of tooling that can help organizations deeply analyze and react to user behavior across every touchpoint with the business.
Challenges with 1st-gen product analytics
The first generation of product analytics was born out of Web 2.0. The tools that emerged (Mixpanel, Amplitude, Pendo, Heap, etc.) have served product-driven companies well by providing top-level visibility into product usage that they didn’t have before. However, in today’s digital landscape, there are several challenges and deficiencies with 1st-gen tools. These include:
1. One-size-fits-all analytics
1st-gen tools are designed to provide out-of-the-box reporting templates for basic self-service reporting. Unlike business intelligence software, which has slice-and-dice capabilities for advanced ad-hoc analytics, product analytics solutions generally offer only pre-defined reports on rigid data models. Moreover data collection and processing is designed to fit this pre-defined data model, and data that’s not needed to produce the prescribed reports is discarded along the way.
What if the user wants to ask a follow-up question, and it’s a domain or business-specific question that goes beyond the reporting templates? The workarounds today are painful. Typically, you’ll need to export all the product telemetry (or at least what telemetry remains from the prescribed data collection process) into a relational data warehouse, and write complex SQL for modeling and analyses. The users of product analytics tools are business users — they likely don’t have the latest skills or access to tooling to do this. As a result, they have to rely on overloaded and organizationally expensive data engineering and data science teams.
2. Siloed single-channel visibility
1st-gen tools typically focus on a single channel of the customer experience: the in-product experience. While that data provides some insights, it’s incomplete. In a multichannel, always connected ecosystem, customers interact with brands not just through their instrumented products, but also through out-of-product channels like social, customer support, marketing, events, and offline channels. Product analytics tools offer little, if any, support for multi-channel analysis.
Product analytics tools also lack the ability to sufficiently enrich product data with data from other business systems, contributing to the fragmentation of analytics solutions in the organization. For example, a tool may offer a well-modeled notion of a “user” based on the events being captured in the product or application. But what is the complete profile of this user? Are they a power user that’s part of an account paying $1M and up for renewal this quarter? Customer engagement or financial systems possess this data. Other examples of business context include supply chain, sales history, or support channels, but such data is not in the event streams for 1st-gen product analytics tools — it lives in other business systems (Salesforce, Zendesk, Marketo) or in internal data lakes/warehouses.
3. Closed legacy stacks
1st-gen tools have built proprietary closed data stores for the event data they collect. This is typically messaged as a necessary step to generate analytic performance. The results in yet another a black-box, siloed data store. Legacy product analytics tools predate modern, open Lakehouse data architectures. Their data is not in industry standard formats, nor is it exposed through standard APIs such as JDBC, and SQL.
As a result, for even a moderately large amount of data or analytical complexity, customers have to build ETL jobs to dump data out of these stores into other databases. They then have to use yet another tool to analyze this data, creating more overhead and increasing the TCO of their analytics machinery.
4. Scale and performance issues
1st-gen tools evolved from basic reporting templates for simple use cases with simple computation engines. They weren’t designed to scale to support the sheer amount of raw data that businesses are capturing today. To avoid performance degradations, these tools summarize and purge large amounts of high-fidelity data that could otherwise provide answers to important business questions.
Next-gen tools help you self-serve answers to the next question
To address the challenges of 1st-gen product analytics tools, a new generation of tools is emerging in the form of operational intelligence (OI) systems. These systems are designed for business users to easily ask more sophisticated questions from a single platform — a platform, that delivers what the business needs to know now, not just last quarter, last week, or yesterday, and unconstrained by channel, function, or system, so that teams can focus on quickly taking action, fixing issues, and/or improving outcomes.
OI systems take product analytics to the next level. Organizations aren’t limited to looking at the behavior of users in the product — they can also see how users behave across other touchpoints, compared to past historical behavior with dynamically evolving cohorts, and more. For example, instead of just measuring the shopping cart abandonment rate inside an app, teams can attribute channels like supply chain, web-to-retail, offline advertising, or customer loyalty to that metric. With these insights from across business systems, they can personalize their user outreach at the point of cart abandonment.
The next generation of tools for understanding how customers interact with your brand can be found in operational intelligence systems. Here are the features to look for:
Rich business context and modeling
Next-generation analytics tools can connect to any business system out-of-the-box, from multi-channel sales and supply chain management platforms to customer engagement and support systems. This includes streaming platforms (like Kafka), static stores (data warehouses, data lakes, RDBMSs, key-value stores), SaaS applications, and more. The benefits are two-fold. Product teams and other stakeholder functions all have a 360-degree view and full context of their operations and customers, regardless of the systems of record in place for Sales, Marketing, Support, Finance, or HR. And this shared visibility facilitates collaboration between these functions as they all work towards the common goal of delighting customers.
In addition to system integrations, these tools offer advanced modeling capabilities. They allow the organization to schematize raw data in various forms into meaningful relational business entities, with multiple views of the data for various business scenarios, support for semi-structured data, and resiliency to changes in the source data.
Automated actions to personalize experiences
Every customer is unique. Companies that recognize this and serve each customer’s unique needs have a competitive advantage. Providing personalized experiences wins customer loyalty and more business. Next-generation tools use insights from behavioral analytics and real-time data processing infrastructure to enable dynamically tailored personalization for every customer, across all customer touchpoints, in real-time.
Business monitoring and alerting
We are in an era of rapid increase in the velocity of business operations, with the COVID-19 pandemic accelerating digital transformation by many years. Every enterprise needs to react to business events faster than ever before to stay relevant, meaning they need to have a real-time pulse on their business operations. Next-gen analytic systems reflect this evolution. They’re capable of processing data in real-time and instantly alerting product and customer-facing teams about issues and opportunities.
Empowered business users
Next-gen tools allow business users to self-serve for data access, modeling, and analytics, with minimal reliance on IT and data engineering teams. No longer confined to prescribed reports or dashboards, business users will have intuitive point-and-click interfaces for accessing, analyzing and visualizing data from any data source. They’ll be able to enrich and extend the organization’s data model according to their domain expertise. And ultimately they will be able to ask more sophisticated questions with advanced analytics.
Convergence of event and state (BI-style) data analytics
Next-gen tools mix traditional product analytics with BI-style advanced dimensional analysis while maintaining a friendly interface for business users.
They can easily combine BI-style state data with any analysis, such as sequences and paths of events. For example, across a dynamic cohort of users who converted within a week of trial, it’s easy to peer into spend by age, region, and product category. With this flexibility, teams can perform the kind of rich ad-hoc exploration, one question naturally leading to the next, that rewards curiosity rather than constraining it.
Next-gen tools go beyond reporting on what has happened to providing a lens on how and why things happened. With these capabilities, analytics can become more forward looking and begin to predict and even to prescribe. Predictive analytics uses machine learning techniques to forecast user behavior, like predicting a user’s propensity to churn based on training data from other users’ behavior. With prescriptive analytics, the system recommends actions to influence user behaviors that positively impact business outcomes, like launching a campaign targeting users that are at risk of churning.
Modern Lakehouse architectures
Data lakes built on stores such as AWS S3 are emerging as the central repositories of all data in modern enterprises. They offer the ability to store massive volumes of diverse data (structured, semi-structured, and unstructured) at a very low cost. They combine enterprise-class reliability, availability, and security with open APIs to make data accessible to any application. To query data directly from the data lake without creating analytics silos, a new architecture called “Lakehouse” has emerged as a best practice for data processing. Next-generation tools are built from the ground up as cloud-native, Lakehouse systems that provide massive cost, scale and manageability advantages.
Take your product analytics to the next level
Are you struggling to overcome the challenges of your current product analytics solution?
- One-size-fits-all analytics
- Siloed single-channel visibility
- Closed legacy stacks
- Scale and performance issues
More importantly, are you expecting more from a modern platform?
- Rich business context and modeling
- Automated actions to personalize experiences
- Business monitoring and alerting
- Empowered business users
- Convergence of event and state (BI-style) data analytics
- Predictive analytics
- Modern Lakehouse architecture
If you answered yes to any of the above, you should be considering a next generation operational intelligence platform to maximize the value of your investment in product instrumentation and begin asking more sophisticated behavioral questions. To speak to an OI expert and learn how to act on deeper behavioral insights, please drop us a line.