construction site

It wasn't too long ago when business users struggled to answer simple questions like, “Who are my top 10 customers? Who are my top five suppliers? Are people opening and reading our email campaigns?” 

As recently as the early 2000s those questions were nearly impossible to answer. The advent of out-of-the-box business intelligence made those questions much easier to answer by managing and delivering static and descriptive analytics. 

But while businesses were able to query (including ad hoc), report and analyze the process of creating these reports, delivering agility throughout the analysis process was difficult, time consuming and didn't address business needs. And, with answers to these initial questions in hand, stakeholders increased the breadth and depth of their inquiries. 

IT picked up the brunt of the demand as data and analysis requests became backlogged. Ultimately, business users began thinking of IT as a bottleneck to accessing data. 

Custom Analytics Solutions Become Custom IT Headaches

Fast forward to today’s surge of data growth and complexity, and the frustration around the analytical process has only heightened.

Modern-day businesses face a typical, but entirely different scenario. Perhaps you are a data-driven business that started out with a data lake built on top of Hadoop that stores your unstructured and structured data. In order to combine all of your data assets you need to have resources, expertise and tools. In general, the DNA of your business dictates that you can, and will, “build” it yourself. 

However, the requirements to build seem limited and costly. For example, Hadoop is not inherently easy to leverage for analytics and modern-day businesses need resources such as a data scientist as well as access to training, tools and resources. In this manner, these businesses build analytic applications on top of Hadoop structures to create insight and opportunity using the raw data found inside and outside the firewall. 

Custom solutions like this “build-focused” infrastructure on top of the data lake quickly turn from an agile and quick approach to deducing insight from your data, to a headache for IT management. Soon, the business starts to question the governance and accuracy of the data and analysis, which becomes a slippery slope. 

This slippery slope leaves modern day businesses in the same situation as enterprises — needing a complement to existing infrastructure that can help manage solutions and deploy self-service analytics. 

Steps to a Modern BI Solution

Both of these scenarios exist in the wild: we see big businesses that invested in early business intelligence solutions and newer medium-sized businesses looking to invest in big data analytics. How can we meet all of their needs at once? The answer is a hybrid solution — build some of that in-house and buy some of this technology so that you can get the best of both worlds.

Throughout my experience with a variety of clients, I have seen the implementation of best practices that incorporate this solution. These best practices will help you make better decisions up front and help the business make more prudent investments. 

Here are few steps to take when you start down the path of building a modern BI platform.

Step 1: Analytic Inventory

Take an inventory of what is happening in your business today to define your business needs. Analytic inventories differ from physical inventory in that you are looking at both physical resources (people, software, hardware) as well as institutional resources (e.g. does your organization look to the data as part of its decision making process?).

Performing an analytic inventory allows the champions of this process to ask the hard business questions and find the pockets of analytic excellence in the business today. This understanding guides the process of near future changes.

For example, I have a customer who wants to move all of their analytics processes to the cloud. Before doing this, their system architect performed an analytic inventory. They discovered that their “health and safety” group delivered analysis and reports on a twice-daily basis. The analysis told the foreman of the accident rating on their job site, how workers performed in various weather conditions and the resolution of accidents that had been reported. 

After the analytic inventory was completed, the company used this group’s approach to analysis and reporting as a model for the rest of the organization to help everyone design and adopt new analytic work streams.

Step 2: Watch Your Analytic Pace

Businesses can’t afford to stroll down the analytic lane — walking toward the goal of becoming a data-driven company is much too slow. Equally untenable is running headlong into buying everything in the analytics store and forcing the business to change immediately. 

But a fast clip — buy some and build some analytics — is the perfect medium ground. 

A business can create rapid prototyping projects with in-house expertise to build some of the custom functionality it requires. Concurrently, IT can procure the vanilla functionality from software vendors. This blended approach is practical as it allows organizations to purchase the more mundane functionality while building the analytic functionality that is unique to the business.

Step 3: Prioritize Analytic Delivery

While investigating and architecting this strategy, spend your investigative energies on understanding the analytic requirements for your stakeholders and colleagues.  Every group requires different levels for each type of analytics. 

Business analytics has been described as an evolution from raw data to “prescriptive” analytics and all the steps in between. Understanding where each group is in the evolution helps businesses prioritize and tailor which analytics should be delivered to which group. 

Help the organization see the journey they want to take and determine delivery priorities. The prioritization will first solve the greatest pains and then deliver the greatest value. Both experiences allow the business to adopt the new analytic paradigm more easily.

Change is a Multi-Year Journey

By describing, articulating and justifying change, business stakeholders will know how to proceed in this multi-year journey. This understanding and collaboration will increase the chances that the project succeeds and that cultural change will be achieved, improving the business in ways that people could only imagine before the journey began.

Taking these three steps into account around the assessment of buying some and building some will help an organization achieve a greater understanding of what and how much they want to do of each.

This is the first article in a series of three. Stay tuned for next month’s article in which I discuss what to look for during the buying and building decision process and what factors to weigh.

Title image 贝莉儿 NG