6 steps to successfully incorporate analytics in any area of your business

Published | Wednesday, March 15th, 2017

You have great software, lots of data and an excited team. What you’re missing is exactly how to incorporate analytics into your organization. Might I recommend the following six-step framework? This framework is designed to allow you to apply analytics in virtually any area of your business, by changing the strategic question.

Step 1: Ask the strategic question

The first step of the analytics development framework is asking a strategic question—what business problem are you trying to solve? This question may be as specific or broad as you like. Questions that are broad in nature will likely require more objectives, data and procedures to get an answer. While questions that are specific will require fewer objectives, data and procedures to get an answer. Once you have the question, you might think of a few more—and that’s ok!

Step 2: Define the objectives

After determining the strategic question, your next step is defining the objectives that will help you answer the strategic question. This step requires you to break down the question and determine what it will take to get an answer. Some questions will result in ten or more objectives, while others may result in just a couple. As you work through the remaining steps of the framework, you may find it necessary to revisit the objectives to add additional items.

Step 3: Obtain Data

Once you determine the strategic question and define the objectives necessary to get an answer, you need to begin gathering the various data elements necessary for you to complete your analysis. You should begin discussions with your IT department very early in the process to determine your options in obtaining data. An ideal set-up would involve open database connectivity connection directly to the database, reducing your reliance on the IT department. If that is not a possibility, work with IT to develop a production schedule for your required files on a routine basis.

There are two data sets needed in every analytic application: the data for analysis and the data for follow-up. The first set includes all tables and data fields necessary to meet your objectives in testing. The second set includes all tables and fields necessary to complete an effective review of the results. You should obtain all of this data at the outset of the project to eliminate the need to supplement results with additional data late in the process—reducing the likelihood of errors in appending supplementary data.

Tip: Think beyond accounting and financial systems. Organizations have an enormous amount of data outside these standard systems—productivity data, email, instant messages, social media, access logs, etc.—and it should all be considered in this step of the analytics framework.

Step 4: Develop procedures

The development of procedures should occur in phases. The ultimate objective for the recurring, or routine, tests is to develop automated analytics. However, that is not the beginning of the process. Procedure development, typically, can be broken down into four phases:

    • Phase 1: Ad Hoc Individual: The first phase of procedure development is simply determining how to accomplish the first procedure. At this point, the process is manual in nature and re-performance requires completing all steps manually. This phase should continue until successful completion of the procedure occurs on a regular basis.
    • Phase 2: Automated Individual: Once the ad hoc individual procedure is regularly successful, it is time to automate the procedure. This step requires coding expertise in the software solution you use. We recommend using a modular approach to automation. This approach uses one script to normalize the required data set and another to perform the testing. We recommend this approach to facilitate maintenance in the event the underlying file structure should change.
    • Phase 3: Automated Groups: Over time, you will accumulate a number of automated individual procedures. We recommend grouping these procedures into groups that accomplish a similar objective or answer a strategic question. The automation at this stage will allow the user to run a group of tests through a single automated interface, rather than having to run them individually.
    • Phase 4: Continuous Analytics: After creating automated groups, the procedures are ready for continuous production. This may not mean real-time application. The speed of the business process will set the definition of continuous. The overarching principle of this phase is the analytics run without human interaction.

Step 5: Analyze results

The penultimate step of the framework is the analysis of results. The first aspect of this step is determining if the results appear reasonable based on your expectations. Along with this, you need to identify potential false positives in the results or false negatives missing from the results. There is an expectation of false positives the first time a procedure runs. If you do not have false positives, the initial focus of the procedures is likely too narrow—this may result in false negatives. As you work through the results, you will find the right balance of false positives and actual findings to have comfort the procedures are effective.

The second aspect of this step is determining if the results meet your objectives and answer your strategic question. If they do not meet the objectives, you should revisit the procedure development step and add further procedures. If they meet the objectives, but do not answer the strategic question, you likely need additional objectives, additional data or better procedures. Your analysis should tie back to the overall goal of the analytics developed and that is to answer the strategic question.

Step 6: Manage results

Once you have results that answer you strategic question you move to the final step—results management. How will the organization use the results of the analysis to drive the success and sustainability of the analytics function? If no one is using the results, the analytics function will fall flat from lack of support. If delivery of results is not timely, the analytics function will struggle to expand due to a lack of confidence from others. I recommend having a plan for how you intend to use the results of the analytics early in the process. Then when you get to this final step, you are ready to put the results to work.

An important consideration in this step is who will be using the results. It is important to recognize there are likely multiple users of the results within the organization and what is important to each of the users may differ. I recommend developing multiple results delivery options and formats, depending on the audience.

By following this 6-step framework, you’ll find the structure you need to successfully incorporate analytics into your organization.


About the author:
Jeremy Clopton is a Director in BKD, LLP’s Forensics & Valuation Services division, leading the Big Data & Analytics and Digital Forensics practices. He has more than 10 years of experience applying data analytics in fraud prevention and detection, risk assessment and business intelligence. He is a frequent speaker on the topics of data analytics and visualization, forensic accounting, data analytics program design and the value data can bring to an organization, including speaking at ACL Connections in 2013, 2015 and 2016


Published with permission from ACL Services

  Get in touch with us!

In compliance with Section 45 of the ECT Act please confirm the following:

I would like to receive future communication from CQS.

Leave a Comment

Your email address will not be published. Required fields are marked *