There are five scenarios: two that typically fail, two that sometimes work partially, and one that has emerged as best. Let’s take a look at each:
1. Helping for solving problems?
This scenario often starts with the CEO (sometimes prompted by the board) deciding to hire a data scientist and establish a data analytics group. The data team sets out in the organization with great aspirations but without specific guidance to find business problems to solve. The data scientists, however, don’t have a practical understanding of the business, and the business leaders don’t know what, exactly, the data analysts are supposed to do or how to use them. As a senior executive of a very large enterprise said, “Our CEO hired a data scientist who reports to me, but I’m not sure what he does or what to do with him.” As the business leaders and the data scientists try to figure out how to relate, not much business value is created.
2. The great passion!
Well-intended enthusiasm for putting data science to use can lead to overly ambitious aspirations to impact the entire company at once. The reality, however, particularly in large companies, is there are too many legacy data systems, too many practical issues, and too few people on the data science team to produce a significant business lift across the whole company in short order. Business results typically fall well short of high expectations. As an executive of a multinational European manufacturing company observed, “We’ve been at this for three years, and spent millions of euros, but we don’t have much to show for it.” In the end, not much business value is actually realized.
3. Big bang!
The third scenario has promise: C-level leaders direct that data analytics should be adopted throughout the company. The practical use is left to the discretion of each business unit leader or function head. While data analytics gets grounded in and kept close to the business, much depends on if and how individual business heads choose to use it. Some embrace it and achieve significant results; others aren’t sure what to do or else avoid it. “Data analytics” often becomes just enhanced business reporting. Databases, systems and tools proliferate. With fragmented efforts, it is difficult to scale the resultant activities and determine how much business value is being created.
4. Three years and $10 million from now, it’s going to be great.
This rational approach is undertaken with all the right intentions — that data analytics can create business value, but require commitment, investment, and time. The problem is this approach typically results more in process than business outcomes. Often it involves a series of workshops, committees, and meetings that drag on without much to show for it. Multiyear investments are difficult to sustain without any business results in the face of competing budget demands and changing business conditions. A large industrial company, for example, has been planning, developing, and discussing their data analytics initiatives for years, but executives wonder where the effort is headed and when it will show business value. Despite a promising start, too much time passes without business results, and support wanes.
5. Begin with high-leverage business problems.
Finally, the approach that works best: Identify a small number of “high-leverage” business problems that are tightly defined, promptly addressable, and will produce evident business value, and then focus on those to show business results. The specific business problem drives the team to identify the data needed and analytics to be used. Quick wins demonstrate business value. For example, a company that operates medical imaging clinics saw a high-leverage problem in patient “no shows.” The company set out to predict and reduce no-shows for the benefit of all involved: patients, doctors and technicians. Reducing “no shows” directly improves the bottom line. There’s no substitute for business results to build credibility for data analytics and sustain commitment.
Some notices for Data Analytics
As we look across these scenarios, best practices become clear, including:
- Data science can’t happen in a silo. It must be tightly integrated into the business organization, operations and processes.
- There needs to be joint prioritization. Business leaders and data scientists should jointly decide which business problems to focus on. If there is any question about priority, the final call should go the business heads.
- Leaders need to be conversant in data science. Business leaders don’t need in-depth expertise in data science, but they require a basic, working understanding. Being conversant enables business leaders to work effectively with their data science teams.
- You may need to accept “inconvenient outcomes.” Data inevitably creates transparency and reveals business insights that can be unexpected, uncomfortable, and unwelcome. Data analytics will unearth inefficiencies and misconceptions that complicate leadership and disrupt conventional thinking. Business leaders who crush or ignore answers they don’t like will rapidly undercut the value of data analytics.
With observing the different approaches taken by a wide range of companies, we can see what works and what doesn’t to connect data analytics to creating real business value. Because if your data analytics isn’t adding real value to the business, it’s not going to be successful or sustainable.