The Five Dimensions of Data Analytics

How to get started using analytics to improve business performance

by Chris Rebant

Data Analytics has been an increasing area of focus in most businesses over the last couple of decades. The analytic process grew increasingly complex, requiring highly skilled quantitative professionals to develop analytics using complex tools. Over the last several years, the technology has evolved to allow business users to perform quantitative analytics that provides excellent insights without the need for the level of quantitative training required as little as 5 – 10 years ago. Quantitative expertise is still needed to provide oversight and develop complex analytics. However, the new software tools have allowed a broader set of users to perform very insightful analytics quicker and more focused on specific business needs.

It is costly and frustrating to spend significant time developing analytics without providing value if there is no clear goal for undertaking the process. Establishing goals does not have to be a complex process and will provide significant value.

Understanding the five levels of analytics is a very helpful way to understand where your particular area may be in the analytic process and provide insight into appropriate steps to move forward in an efficient manner. Remember that this is an evolving process that will change as the business needs change, knowledge expands, and additional techniques/technologies are added. I have always worked with the expectation that if I am still doing everything the same way for 6-12 months, then we are not expanding, improving the process, and highly likely not meeting the business needs.

Descriptive Analytics is the first step in the analytics process. It helps stakeholders understand the current outcomes of their business or, more simply, states what happened.

What are the key components of Descriptive Analytics?

Collection and Storage – Gather relevant data to process, analyze and visualize. This is also where data governance should begin to develop to ensure consistent and accurate information is leveraged and managed.
Summarization – Establish the practice of summarizing large data sets into useful information focusing on that which is relevant to the business needs.
Business Value – Effectively describe outcomes to stakeholders.
Performance – Develop key performance indicators (KPIs).
Metrics – Define specialized metrics to track performance in particular focus areas of the business to ensure stakeholders understand, support, and use the metrics.

Discussions with the stakeholders to determine their needs will help develop the plan for beginning the analytics process. This will improve efficiency, focus the efforts, and will ensure timely and strategic decision-making that supports positive results and growth.

These suggestions will help build a strong baseline for the ongoing development of the analytic process.

Measure what matters. Know what metrics are important to the business partners and clearly define them.
Streamline reporting requests. Create overall KPI reports and dashboards that are readily available within an agreed-upon timeframe (daily, weekly, monthly).
Ensure consistency through templates. Create master reports for users to customize and/or create their own ad hoc reports using the identified data. Multiple tools are available for this type of work, such as Qlik, Tableau, and PowerBI. This is a topic for a future blog!
Test reporting efficacy and value. Review results with users to determine the value of the provided analytics, identify opportunities for improvement, determine the accuracy of the underlying data, and develop a working relationship with your business partners.

Diagnostic Analytics is the next step in the analytics process. The focus here is on the “why” or how we got here.

Root Cause Analysis – Determine the why behind the current issues.
Performance Indicators – Evaluate the success of each activity. Did the initial performance indicators provide useful information? Are there other metrics that would be helpful?
Trend Discovery – Determine improvement and degradation trends.
Outlier Analysis – Identify anomalies in the data.
Statistical Techniques – Find relationships and trends that explain the anomalies.

Be open-minded about the results of the analysis. It may not be what was expected or hoped to see. However, it may give you a deeper understanding of the business drivers, potential data issues, or new outcomes. It may identify process changes that could affect your outcomes. Changes in results can be early indicators of changes in the business, such as customer needs, new competition, or even inefficient processes or issues.
Maintain a regular cadence of analysis to discover trends sooner than later. Regular analytic review sets expectations for the analytics, builds trust in the process, develops the business partnership, and broadens the knowledge of all who are involved.
Treat anomalies as opportunities. They can point to data quality and influence changes that need to be made or are simply an exception that is not relevant. It also can be an early indicator of opportunities or challenges!

Predictive Analytics is the third step in the process. This leverages the results of the descriptive and diagnostic process to begin predicting future trends. It is important to have enough historical and real-time data to be relevant and useful. Here are some ways you can predict future trends.

Proactive Decision Making – Leverage insights into what may happen in the future.
Predictive Process – Utilize techniques using historical data to identify trends and predict future behaviors.
Analytics Technology – Enable predictive analytics tools.
Predictive Techniques – Leverage statistical and machine learning strategies.
Predictive Neural Networks – Follow decision trees and regressions to predict behaviors.

Predictive analysis is a continuous cycle. The steps below will help you improve your process to anticipate trends and behaviors better.

Start by defining why you need the data analysis. What are you trying to predict? The “why” of what you are trying to do determines what you need to achieve a targeted outcome and if the targets are achievable. Business partners provide great feedback in this process and can provide interesting insights based on their experience.
Collect historical and real-time data to compare and analyze. Looking at the historical information and results helps set a baseline for your predictions. (e.g. What are the core drivers of the results, and what happens when the drivers are changed?) A significant portion of this information would have been provided in the Descriptive and Diagnostic steps, hence the evolutionary steps. Additional information is typically added, and some are deleted as the process evolves.
Interpret the results to predict what may happen in the future. At first, limit the scope and then begin to expand the scenarios leveraging additional data and assumptions to develop multiple outcomes as your comfort with the numbers and process grows.
Keep a manageable timeframe to run the process, including predictive analytics. Keep the timeframe within the capabilities of the team aligned with the timing and expectations of the business to ensure you meet their needs and keep the focus on what is important. This is very important as it is easy to get lost in the details, remember the primary purpose is to inform the business, and it is an evolving process.
Be flexible and continue iterating to enable meaningful predictive analytics. Continual review and testing ensure the accuracy, functionality, and usefulness of the analytics.

Prescriptive Analytics is the fourth step in the analytics process. This will leverage what was learned in the first 3 steps to begin to make recommendations for specific actions to meet the company’s goals and objectives. What are some key considerations?

Action Drivers – Determine key future actions and what should be done to improve outcomes.
Certainty Drivers – Inform decisions with insights in the face of uncertainty.
Pattern Identification – Leverage techniques and machine learning strategies to find patterns in large datasets.
Forecasting – Analyze past decisions and events to determine the likelihood of similar and different outcomes in the future.

Prescriptive analytics has been the target environment for many business analytics journeys, and the continual expansion of tools capabilities has allowed businesses to begin achieving this very desirable stage of analytics! Understanding the options uncovered will suggest the best course of action given expected environmental changes that will reduce risk, improve performance or support changing direction. It is important that the first three steps are complete, including documentation, ongoing monitoring, and business acceptance. Here are a few suggestions to develop or improve your company’s prescriptive analytics process.

Automate your targeted outreach. Plan, create, and oversee automation flows to tailor relevant messaging to your target customers based on their motivation and needs.
Analyze the various outputs. If you use a third-party analytics tool, consider comparing the prescriptive results to the results from descriptive and predictive as a validation technique.
Evaluate your findings and potential actions to determine outcomes and how they impact your goals. This type of analytics will identify changes in the underlying business to reveal opportunities to adjust/ change existing processes and products, identify the impact of unexpected changes or identify new offerings.

Artificial Intelligence (AI) is the final step of the analytics process. The focus of this area has been to automate repetitive tasks to free up time and improve productivity (e.g. general ledger entries, including assigning approvals, performing additional entries, and creating payments). This step has been growing as the technology and methodology have evolved to improve ease of use and dramatic increase in the speed and ability to access very large datasets from various sources. This has tremendous capabilities, as we have seen in recent months, and also can add risk given the potential to add bias inadvertently.

Perception – Acquire, interpret, select, and organize sensory information.
Visual perception – Directly map input into output via a learning process.
Speech recognition – Create a front-end of a system that can perceive and understand spoken language.
Automated decision making – Leverage data into a single, unified view to automate actions.
Translation between languages – Property translates your content by interpreting the intent of the source content.

AI is having huge impacts across a number of repeatable tasks in many businesses, including customer interactions, ledgers (as mentioned above), and other items. It is a rapidly evolving process as the capabilities continue to improve. Here are some suggestions to help improve your AI Analytics journey.

Review your current processes and activities from end to end. Determine which tasks, typically repetitive, your team can automate to allow assigned team members to focus their time and energy on strategic efforts and leverage their areas of expertise.
Test your current state technologies and processes as related to future goals. Review automation systems for potential limitations or outdated processes.
Ensure that AI is combined with human evaluation and oversight for optimal effectiveness. Consider using ongoing reviews, accessible reporting, and appropriate approvals embedded in the process.
Think, engage staff, and design solutions cross-functionality. Build a diverse team from different departments to ensure an appropriate level of expertise and responsibility are represented. Hence, the strategy meets objectives for the key internal stakeholders.
Ensure your metrics are strategic and tactical. Align your AI strategy with your key performance indicators (KPIs) to measure success directly to your target market, whether internal or external. Continue to review, iterate, monitor, and update the AI processes.

For more information on how Five to Flow helps organizations accelerate change and growth, feel free to contact us. You can visit our Analytics Solutions learn more about our areas of expertise.