Beyond Gut Instinct: Improving ROI with Data Driven Decision Making
Data is the raw material of decision making. Businesses that use return on investment (ROI) and similar metrics need timely and accurate data to make decisions. Unfortunately, many businesses are not even attempting to collect data. A recent Harvard Business Review article noted that 69 percent of respondents reported that they have not created a data-driven organization, and 53 percent state that they are not yet treating data as a business asset.
Companies that are keen on maintaining a competitive edge are embracing a "data first" framework for making mission-critical decisions. For example, in the early stages of evaluating a proposed action — such as implementing new functionality in a web application — a team should determine what data it will generate, how that data will be collected and analyzed, and how to measure success using key performance indicators. Data collected from sources such as clicks or customer engagement can be aggregated and sorted using Google Analytics. From there, it can be compared to key performance indicators to help measure success.
Automating Data Collection
Shifting to a data-driven culture begins with implementing practices that facilitate the use of analytics. Organizations need to start collecting metrics and evaluating decisions in light of those metrics. Initially, data collection may absorb a significant percentage of the time and effort required for data-driven decision-making. But once a data collection process is established and automated, more time will be available for more informative analytics.
Those analytics practices enable decision makers to use data to evaluate multiple options. Take, for example, the ubiquitous metric, return on investment (ROI). Knowing the ROI on one investment option is not enough, it should be compared to the ROI of other options as well. Otherwise, an organization risks making decisions with a poor understanding of the context of the decision as well as alternatives that may be a better investment.
"Absent robust analytics, the most basic measurements are guiding the business decision-making – and often not in the right direction. This leads to focusing on the lowest cost per customer acquisition instead of stopping and asking if the customers we are acquiring are, in fact, the right customers to begin with" said Michael Loban, CMO of InfoTrust.
It also is important to use multiple metrics for a more complete understanding of domain under investigation. For example, metrics, such as customer lifetime value (CLV) which takes into account past purchases and behaviors, provides more informative data for decision makers than a more limited measure, such as monthly average purchases. For example, scoring a customer value based on a moving average of purchases over the past three months can easily miss seasonal variations in a customers purchasing behavior.
"CLV is the single most important metric for measuring gross profit and success over time," observed Johannes Tarnow, an e-commerce marketing analyst. "Knowing the CLV of a customer will help you to strike the ideal balance between customer retention and acquisition."
CLV is less prone to variation due to seasonal trends or unusual events that disrupt business, such as extreme weather events. Monthly data is still valuable though. When analyzing data that may be seasonal in nature, such as monthly retail sales, decision makers can compare current month sales with the sales from the same month in previous year.
Take game maker Electronic Arts for example. They focus on calculating CLV for customers and then focusing on the most valuable of those customers. The data they collect is used to help develop more products targeted to high value customers.According to Peter Fader, marketing professor at The Wharton School, "They find out which games these people are playing, and which scenes within the games they are staying the longest on. Electronic Arts uses this information for creative copy ... and they also share it with the game developers, so that they can come up with cool games to appeal to these people."
Another step in a data-first organization is evaluating the outcomes of decisions. Results must be transparent. Everyone involved should have access to dashboards that reflect the outcomes of these data-driven decisions as well as the key performance indicators.
The final step of the data first approach is getting the analytic results to decision makers. "You can use traditional analytics such as dash-boarding," says Scott Hebner, vice-president of marketing at IBM Data and AI "You have to operationalize all of this, because ultimately the analytical or AI models have to get to the people who are doing all of the work."
Also, keep in mind that some of the most useful data for your analysis may not be generated internally. For example, aggregated, anonymized payroll data, coupled with internal employee data, can help human resources professionals better understand market trends in order to formulate competitive compensation packages. In addition to proprietary data sources, there are many open source data sets useful for business analytics available at Data.gov, which has data about U.S. business.
Data-driven decision making is not an advanced technology available to a limited number of organizations. Advances in analytics software and data collection tools have effectively democratized the tools organizations needed to develop a data first culture. What is needed to realize the benefits of a data-driven culture is a commitment to automating everything from data collection and analysis to bench-marking and ROI measurements. Knowing what your success metrics are will you give you a head-start when it comes to evaluating which business strategies are working best for your company.
Keyence has over 40 years of experience with data driven decision making. Download the 4 Step Automated Process digital brochure and not only learn more about data driven decision making but the tools to support it.
LATEST POST : Driving into the unknown of analytics (Part 2) – The Model T of analytics