Mike Romeri, CEO . Download pdf version
All organizations—whether big or small—need data and advanced analytics to improve their business decisions. Why? Because organizations applying analytic techniques outperform those using ad-hoc decision-making methods by at least 10%, initially. So, where do you start, and how can you prepare your business for a big-data analytics project?
Starting a Project
Before you jump in, it is important to understand what you need to start an advanced analytics project. Uncovering hidden relationships within your data that you didn’t otherwise know about is very interesting, but is that enough? Does it help you make better business decisions? It may very well help you with decision-making in some cases, but—in the framework of a business strategy—that approach is called a fishing expedition. A truly rigorous scientific process involves asking a question, building an experimental model to answer that question, testing one variable at a time, and then drawing conclusions based on the evidence. Therefore, it's advisable to start every advanced analytics project with a well-defined question or use-case.
Data scientists need a business question to start any data-analytics project. They need to build a model to test. This model will include the data that should be tested, the types of algorithms capable of answering particular types of questions, the requirements of your business, and the parameters within which your business operates.
What Results You Can Expect
It is also important to understand what you will get out of an advanced analytics project even before you begin to frame your question. A priori, the ability to predict the outcome of a decision before the decision is made gives visibility into the future and can allow the best solution to be selected. The use of advanced analytics to predict the outcome of business-decision alternatives before decisions are made has a huge business advantage. To be able to go from a handful of intuitively identified alternatives to thousands of systematically defined decision alternatives can be transformative for your business. Advanced analytics allows you to do this and more.
Predictions on what will happen in the future can leave the end-user wondering, what do I do now? In real life, there are typically many possible directions one can go, given a single prediction. Advanced analytics can address this problem by identifying the best recommended action for the business user, one decision at a time, using internal and external data along with a library of mathematic algorithms to test thousands of “what if” scenarios. The “what if” scenarios can total in the thousands and are based on your business use-case and allowable parameters such as compliance rules and more.
The optimized decision recommendations result from testing algorithms until a “best-fit” algorithm is found. Best fit is defined by your business expectations, requirements, and parameters. The best-fit algorithm is then used as the basis for testing all of the what-if scenarios. Thousands of iterations of potential outcomes done in milliseconds, followed by the identification of an optimized recommended action, comprise what is known as prescriptive analytics __ the prescription for the best decision you can make. Having a calculated, optimized, prescribed action delivered to the end-user quickly and at the point of use is referred to as “augmented analytics” by Gartner.1
4 Key Questions to Answer
Before diving into whether or not you need augmented analytics to stay competitive in your business, you should apply some basic principles, including a methodology that focuses on business priorities and the involvement of all stakeholders and experts—within and outside the company. So, what are the business-layer issues you may want to consider? Here are some basic questions:
- Which key performance areas should I focus on?
- What needs to be optimized for each key performance area?
- How will my business process change?
- How will new processes using analytics be adopted?
With these questions in mind, you are ready to look at how data and analytics can help translate your business strategy into value and begin to develop your Analytics Roadmap. Although these questions may seem fairly straightforward, we often find that stakeholders have different priorities, different levels of understanding of the business, and different short- and long-term goals __ all complicating the development of a viable analytic initiative.
An Analytics Roadmap specific to your business can guide you through the requirements you will need before you can go to the next steps. It helps you to prioritize which key performance areas you should address first and second, based on business-stakeholder knowledge and, importantly, data-science knowledge of where analytics can truly bring value. It is a myth that applying data analytics to any business question will improve outcomes. The process requires both technical and business expertise. It is often helpful to have an unbiased view from an outside source to help you develop the best Analytics Roadmap.
Creating and using data models is vital during the development of your Analytics Roadmap. Data models refer to identifying what data are available, what data are useful, and what data will help to improve specific business decisions. You can’t build an analytics strategy without understanding what data you have and what you will need. Well-developed data models can be the basis for formulating potential analytics initiatives, identifying potential drivers of the business, setting analytic priorities, and achieving optimization objectives for the business. Armed with insights gleaned from the analytics strategy process and the set of data-models that are generated, companies can then move on to the technology considerations that enable them to capitalize on new analytical capabilities.
4 Key Technology Areas of Expertise
There are four areas of expertise companies either need to assemble in-house or acquire from outside to effectively use analytics:
- Data Management
- Data collection and cleansing
- Data integration, mashing, tagging, condensing
- Information management
- Scaling and security
- Physical storage and cloud options
- Executive dashboards for KPIs
- Granular drill-down
- Real-time transactional
- Sharing and collaboration
- Data Science
- From simplest to most sophisticated
- In-house versus outsource
- Scale, variety, and complexity
- From concept to production
- Enabling business processes and downstream business applications
- Soliciting feedback
- Operating models and governance
The final outcome of all of these efforts is the creation of a comprehensive plan: your Analytic Roadmap. It will contain analytical components that are based on a multidimensional, sequential project plan. Each phase will detail new implementations of the platform and technologies; data and governance; skills and capabilities, and business outcomes. Each paves the way for your company’s transformation to a digitally empowered business!
Augmented analytics is the next wave of disruption in the data and analytics market. It leverages ML/AI techniques to transform how analytics content is developed, consumed, and shared. Data and analytics leaders should plan to adopt augmented analytics as platform capabilities mature.
- Gartner, “Augmented Analytics Is the Future of Data and Analytics,” by Analysts Rita Sallam, Cindi Howson, and Carlie Idoine (ID G00375087, published October 31, 2018)