12 Ways To Connect Data Analytics To Business Outcomes
Even though more organizations are attempting to become data-driven, many of them still aren't able to link data analytics to business outcomes. Some of the challenges are obvious. Others aren't. Here are our tips for avoiding the common pitfalls.
![](https://eu-images.contentstack.com/v3/assets/blt69509c9116440be8/blt31130e089ac5811e/64cb45770b91a9b85405ed8f/1-ball-72374_1280.jpg?width=700&auto=webp&quality=80&disable=upscale)
One could argue the purpose of data analytics has always been to achieve business outcomes. Yet, enterprises still struggle to realize the potential business value of their investments. Despite the availability of a wide array of improved technologies, it's easy for company cultures, organizational structures, and even problem-solving approaches to get in the way.
"The fundamental premise is it's a technology problem. It reminds me of the early Internet days [when people said] 'We have this capability, what problem can we solve?'" said Jeff McMillan, managing director at Credit Suisse. "That's not how it works. You have a business issue and need to bring a set of capabilities to bear."
Departmental barriers continue to impede progress. Some companies are restructuring to compete more effectively in the digital economy, but the expanding C-suite may frustrate the ability to drive business outcomes.
"Not too long ago, we had a CIO and a group of people who were the caretakers of the databases," said Anthony Scriffignano, VP and chief data scientist at Dun & Bradstreet, in an interview. "Now you've got a CDO, CISO, CIO, CMO … a lot of C-level people in the room, all with different mandates, reporting to different lines in the organization that should own the data. The reality is that data is like oil. All of these people will only see the value in the oil if they cooperate and make something that people want to buy or the company needs."
The aspiration is there, according to a Forrester Research survey of enterprise architects that was published in October 2015. The challenges arise in translating the analytical results into measurable business actions.
"Technology is way ahead of culture. The vendors packaging data and analytics tools and insight platforms are removing a lot of the complexity," said Brian Hopkins, VP and principal analyst at Forrester, in an interview. "The problem is, you still have to fund it, you still have to get people to agree with it, and you have to overcome organizational issues."
Here are 12 things organizations can do to drive more business value from data. Once you've reviewed these, let us know if any have worked for you. Is your organization challenged to connect data analytics to business outcomes? Tell us all about it in the comments section below.
Does your company offer the most rewarding place to work in IT? Do you know of an organization that stands out from the pack when it comes to how IT workers are treated? Make your voice heard. Submit your entry now for InformationWeek's People's Choice Award. Full details and a submission form can be found here.
A wide variety of improved tools are constantly hitting the market. It's tempting to invest in tools and specialized resources without considering the endgame first.
"Any conversation that starts with tools is the wrong conversation. You should start with a business problem," said Anthony Scriffignano, VP and chief data scientist at Dun & Bradstreet.
One of the biggest challenges faced by Matt Cornett, a data scientist with Sindeo, is getting the context he needs to provide sound analysis back to the business.
"It's not enough to look at the data and figure out what the story is. You have to put it in the context of everything else that's going on," Cornett said in an interview. "In some of the larger organizations I've worked for, if data science isn't part of the business practice, we don't always get the context we need, so it's harder to give [the business] what they need."
Data teams reside in different places in different organizations based on different theories. One theory is that the data teams belong in IT because IT manages the data. Another theory is that the data teams should reside in the lines of business that own the data. Others think the data team should be a centralized service organization, like IT or HR. Another group advocates bi-modal data teams that benefit from centralized collaboration while providing tailored support (perhaps a dedicated person) to individual business units or departments.
"We were speaking with the CIO of a $1 billion company that was debating whether to put its analytics teams in IT or business, and if in business, then where? Finance? Supply chain?" asked Guy Yehiav, CEO of prescriptive analytics vendor Profitect, in an interview. "It has to be a hybrid, because otherwise it alienates the other organizations."
The gap between business units or departments and the people who really understand analytics can be significant. In addition to considering the organizational structure that will best facilitate data-informed business outcomes, organizations should also embrace a collaborative approach to problem solving.
"Companies like Amazon and Google have been able to break down organizational barriers to better understand what granularity of data is available. They develop theories like, 'If I turn this knob or try this campaign or adjust my product, I should see [X] if it's working,'" said Jeffrey Kelly, principal engineer at online travel site Expedia. "On the technology side, there's an understanding that I can't just do whatever I want. I'm here to help achieve a business outcome."
The definition of an ideal business outcome may change as business priorities, markets, customer expectations, and industry standards shift. To keep pace with the rapid rate of change, organizations should consider agile ways of working.
"Our model of core competencies is shifting because the nature of business is becoming more virtual, more mobile, more ethereal. You have the internal structure evolving, and you have this moving marketplace," said David Crais, a certified medical practice executive and business analyst, in an interview. "If I had a medical practice 10 years ago, my success would be defined by the number of patients I see and the level of reimbursement I get. Now, I have patient satisfaction surveys. If my patient is readmitted in 30 days, I get dinged."
Organizations have purchased business intelligence (BI) and analytics tools, hoping to democratize their use. But hope does not drive user adoption. Other organizations have mandated the use of such tools with varying degrees of success. Some companies are restructuring incentives with the goal of modifying employee behavior in a positive way.
Real estate and hospitality company The Cliffs wanted to make sure that its employees understood how its new Domo BI application would benefit the company. It also tailored its presentations so individual departments and employees could understand the benefits in context, according to Jamie Adams, CIO of The Cliffs, in an interview.
Credit Suisse believes in organic adoption. It allows employees to measure their own performance and make recommendations for improvement three months before the same information is used objectively for management purposes. Credit Suisse managing director Jeff McMillan said that that approach is considerably more effective than telling employees every phone call and every digital communication will be monitored to determine whether they're doing their jobs effectively.
Two people can look at the same analysis and come up with completely different conclusions. Similarly, experts and neophytes tend to look at data differently.
"Analytically, what I see as a big miss is a lot of casual observation made with first-order observation: I looked at this data and it says blah," said Dun & Bradstreet chief data scientist Anthony Scriffignano. "It's the second- and third-order observation: I looked at the data which implies blah and I'm going to measure that to see if that is, in fact, happening."
Sometimes, organizations start with the wrong questions. For example, a Dun & Bradstreet client wanted to know how many businesses would exist in China in two years for sales and operations planning purposes. Because the client sold freight insurance, the number of new hotels and restaurants was irrelevant. What it really needed to understand was the number of companies manufacturing products and related details, Scriffignano said.
As companies mature and everything changes -- technology, markets, industry standards, and customer expectations -- the nature of questions may change. Traditionally, companies have been trying to identify their best customers. With predictive capabilities, they may want to know who their best customers will be by the time they launch a campaign.
"We used to want to know, if we changed something on our website, whether sales went up or down. Today, businesses need to know much smaller things, like 'Did I see an uptick in sales in Brazil on the weekends when I ran my advertising campaign targeted to a specific demographic?'" said Expedia principal engineer Jeffrey Kelly.
Some organizations overinvest in tools and human capital, hoping those investments will expedite success. However, it may not be apparent what tools and human capital the organization actually needs until the limits of what exists becomes obvious.
"[When] people say they're going to create a data team, or build a data warehouse, or bring in all these tools, they're solving it from the wrong side of the equation," said Jeff McMillan, managing director at Credit Suisse. "You have to start with a business problem, figure out what's going on, fix that problem, and demonstrate your ability to actually achieve an outcome. Then sit down and figure out what you did right, what you did wrong, and go back and do it again."
The shortage of data scientists and other data team members is fueling the proliferation of easy-to-use, self-service tools for the masses, many of which are now leveraging machine learning and artificial intelligence. While the demand for specialized skill sets isn't going away anytime soon, the tools enable business professionals to get fast answers to questions that historically required specialized assistance. A side benefit of these tools may be that the data team can focus its efforts on the really hard problems.
"With big data, you may have had four to five data scientists who were very capable, but the amount of throughput you could get from them was small. I think the amount of analysis that can be done, and the fact it's being done and being driven by people who are making decisions every day, is really shifting the value of what's being created," said Jeff McMillan.
Failure traditionally has been considered a bad thing, but disruptive companies and R&D organizations realize that a certain percentage of failures will likely be necessary to achieve success. Rather than betting big and losing big (which has traditionally been the case), organizations are betting small, failing fast, learning from their mistakes, and leveraging small successes.
"It's important to be able to fail. The businesses that really go to the next level understand that you will fail a lot, no matter what you do, and that failure is part of getting to success," said Travelocity's Jeffrey Kelly. "To get to the gems, you have to do small things to get to the big things."
Data scientists find themselves in uncomfortable situations when C-level decision makers force them to use data in non-scientific ways.
"The challenge in a traditional business environment is they hire data scientists to validate C-level decisions," said Mert Bay, principal data scientist at marketing analytics platform provider Conversion Logic, in an interview. On the plus side, he said, he's starting to see companies using data scientists and analytics for innovation, "such as figuring out a way to look at things differently or optimizing performance, instead of just measuring performance. It's more difficult in Fortune 500 companies, but a lot of midsized e-commerce companies are getting more data-oriented in their actions."
Data scientists find themselves in uncomfortable situations when C-level decision makers force them to use data in non-scientific ways.
"The challenge in a traditional business environment is they hire data scientists to validate C-level decisions," said Mert Bay, principal data scientist at marketing analytics platform provider Conversion Logic, in an interview. On the plus side, he said, he's starting to see companies using data scientists and analytics for innovation, "such as figuring out a way to look at things differently or optimizing performance, instead of just measuring performance. It's more difficult in Fortune 500 companies, but a lot of midsized e-commerce companies are getting more data-oriented in their actions."
-
About the Author(s)
You May Also Like