Once you’ve established your destination, it’s time to hit the road. An organization’s analytic journey typically progresses through four stages of maturity, as follows:
Each of the four stages requires different tools and processes, and produces different outputs. Understanding these characteristics helps you set appropriate expectations within the organization.
As you plan your analytics journey, consider not just your starting point, but also where you may evolve to mid-term and long-term. Looking at the bigger picture will help you create a program that works today and grows with you for tomorrow.
In our last post we talked about that universal road trip question: “Are we there yet?” Timing is one of the most frequent questions analytics teams face. Unfortunately, there’s no silver bullet that can deliver operational efficiencies, revenue growth, improved customer experiences, greater innovation or stronger compliance overnight. Reaching your destination is truly a multiyear expedition. But, if done correctly, you’ll see business value along the way.
It’s also important to note that your analytics journey may not be linear in its progression. It can start in different places for each company, and even for teams within the organization.
For example, we’ve seen organizations launch data platforms exclusively to develop predictive models, skipping the “See” stage entirely (more on that below). Others may immediately leverage analytics in the “Create” stage to fuel new applications.
Most common, however, is a fragmented journey. Often, various stakeholders, departments or lines of business within the company are operating at different stages of maturity. Some are barely past the starting line; others are well on their way to their destination.
For the analytics team, this makes a strong data platform and a clear strategy more important than ever. Both provide mechanisms to simultaneously support different business users who likely have very different data needs.
Let’s look at each of the four analytics maturity stages in greater detail.
Most organizations begin their analytics journey with the simple desire to “See” their business. Capabilities at this stage usually focus on comparing current conditions to past performance.
In a traditional environment, these implementations often feature static reports and basic dashboards that provide insights from a single data source. When data from multiple data sources is used, the process of integrating data is often manual (and painful). While these analytic tools and methods get the job done, their utility is far from optimal.
One of the biggest drawbacks in a traditional data warehouse environment is the need for specialized operators to create reports due to the complex skills required to access, organize and format the data.
Business users must request reports or analytics, sometimes waiting weeks to receive the final output, and even longer if changes are required. In today’s near real-time world, these delays leave much to be desired. They can even introduce unnecessary risk for business decision-making.
Working only from traditional data warehouses creates other hurdles. As SaaS systems proliferate, growing volumes of data reside outside the data warehouse. This forces business users to accept analytics based on limited data, or cobble together data from multiple sources using spreadsheets and manual methods. Neither option represents a true solution.
Instead, forward-thinking organizations at the “See” stage of maturity embrace self-service analytics. Self-service analytics empower the business to use the tools of their choice and access any or all data for exploration. At Pythian, we call this “bring your own analytics” (BYOA).
A BYOA approach offers many benefits:
Imagine a fitness center with multiple locations. Their analytics stakeholders range from the C-suite to the marketing team and include sophisticated power users as well as far less experienced individuals. With a BYOA approach, everyone can access company data on their own schedule, using their preferred tools:
A BYOA approach serves both power users and those with less skill and experience. It allows them to access the analytic datasets using tools that match their comfort level. For simple analysis, a prebuilt dashboard with a few selection parameters provides ample value. For more advanced users, off-the-shelf SQL tools yield a richer level of exploration. And for data scientists, cloud services that support machine learning models are critical tools.
Key to making a BYOA environment successful is optimizing the data platform that integrates, cleans, organizes and provides access to all the data. The platform should be able to leverage data from an almost unlimited number of data sources, at any volume and velocity.
A strong data platform:
Once business users gain visibility to analytics on present and past performance, predicting the future becomes the next logical stage in the analytics maturity journey. Here, analytics should facilitate forward-looking recommendations that will help the business determine the next steps to achieve its goals.
Predictive analytics can take many forms, and the ability to customize them to specific business needs is powerful. A couple of examples include:
Machine learning technology typically underpins this phase, although organizations can also apply other advanced analytics techniques for predictive analytics. In either case, one of the key requirements for effective machine learning models is significant quantities of good data. This makes the design of the underlying data platform essential to success.
The data platform is the place where data is gathered, stored and organized. It must have a structure that’s both robust and flexible, in order to support insight generation at two different ends of the data spectrum:
In today’s world, a data platform lives in the cloud and features a number of components, including a data lake and a cloud-native data warehouse. Business users will likely need access to governed data in the cloud data warehouse, while data scientists, with their machine learning models, and most applications will use both governed data in the data warehouse and ungoverned or lightly governed data in a data lake. At this stage in the maturity journey, easy access to the right data and ample computing power to run models against it are both important capabilities.
Two other elements to prioritize at this stage are choosing the best machine learning model and automating as many of the supporting steps as possible.
Most models follow a more complicated process than simply accessing data for insights. Failure to automate as many of these steps as possible can jeopardize the return on the machine learning program.
The third phase of analytics maturity leverages analytics to “Do” something within a larger system or application. This is a different mindset than the “See” and “Predict” phases, where generating the insights from the data completes the process. In this stage, it’s all about making analytics actionable. The outputs themselves serve as raw material that feeds into another system or application to drive business outcomes.
A good example is automatically feeding data into a recommendation engine associated with an e-commerce platform. The data is used to predict and enable real-time delivery of personalized marketing offers when a consumer browses a retailer’s website, or helps optimize advertising spend by determining which customer segments see which offers.
For the analytics team, this stage of maturity comes with new expectations for performance and reliability. Using analytics to fuel other applications transforms the analytics program—and the underlying data platform—into a mission-critical business system. That means optimizing for near real-time responsiveness, high uptime and clear data quality.
After all, it’s one thing for a system that produces reports to respond slowly or be unavailable for a period of time. A system whose data is integrated into key customer, sales or revenue platforms, on the other hand, must be robust and readily available.
While companies of all sizes and industries realize analytics yield great business value, they often view the project as an expense or a cost of doing business. It’s common to overlook the revenue potential and product development opportunities you can also derive from analytics insights. Tapping into that potential is the fourth and final stage of analytics maturity.
Here, the product development team leverages the data and analytics previously used for insights and predictions to drive new innovation, such as a new service, or an enhanced feature within an existing product.
A second opportunity in the Create stage is to position the data or analytics themselves as a product that customers are willing to buy. In this case, customers gain access to some or all of the dynamic knowledge collected. This can be through static reports, basic dashboards or even licensing of full datasets.
Often, these new or enhanced solutions come as a by-product of analytics and data work in earlier stages of the maturity journey. Savvy organizations that create feedback loops can then recognize and channel the potential analytic ideas to the right business teams.
Examples of these potential analytic products include:
As these real-world solutions show, the opportunities for new product development are numerous, and limited only by the insights available in company data.
The next (and final) instalment in our series addresses the importance of a modern cloud platform foundation. We’ll also share some practical tips for moving forward. If you have thoughts on any of the posts to date, or on our next planned topic, please feel free to share them in the comments.
For more information, download our latest white paper: Accelerating Your Analytics Journey.
Be sure to read the other posts in this series: