Data science is a powerful tool for business. With a number of practical uses and successful use cases in companies from Amazon to P&G, it’s hard deny the impact of data science methodologies on the bottom line. 20 years ago, that would have led to a long run as a dominant toolset with adoption slowly reaching even the smallest businesses. That was the age of incremental, linear change and it’s no longer the reality for companies in competitive markets.
The pace of change is now exponential. Startups have leapfrogged major players with small data science teams and data driven business models. Companies from tech giants to finance to retail are saying, “Data science was great, but that was five years ago. What’s next?”
The next leap forward is predictive analytics.
With change coming so quickly, it’s easy to fall into firefighter mode. That’s led to some of the most public failures with data science driven analytics. Managing exponential change means preparing for it even when it feels like there’s no time.
Here are three ways your business can prepare for the next disruption and competitive advantage… Predictive analytics.
The business is going to react to this change like it has to every other change; with a thousand why’s. As with any change, success starts with trust. Predictive analytics needs evangelists to build the kind of trust that’ll lead to business wide adoption.
The biggest barrier to building trust in a new technology is fear and uncertainty. The most common fears are:
The most important element of overcoming these fears is a personalized message. That means leadership from the c-suite to supervisors and team leads need to become evangelists for predictive analytics. If employees don’t understand how their jobs are going to be impacted and their contribution to business outcomes improved, they’re not going to get on board.
The way an organization perceives predictive analytics is an indicator of how effective it will be.
A company’s first few predictive initiatives will be aimed at strategy planning and marketing so it’s easy to evangelize the benefits to just these divisions. For predictive analytics to be successful past the first few initiatives, for it to really have legs, initiatives must expand from there. Otherwise, predictive analytics becomes a tool that’s just for marketing or just for the big wigs and that dramatically reduces its ROI.
A good evangelization strategy leads to organic uses of predictive technologies. I know the concepts are taking hold when I start getting pulled aside after meetings with questions that start with, “Hey, do you think we could use that for…” One of those questions from a client’s VP of Operations led to an 18% reduction in supply chain management costs. When everyone in the business starts thinking about how predictive analytics can improve their organization, it results in big wins for the company. That only happens in businesses that evangelize ‘why’.
My favorite story from the data science boom comes from a sales engineer at one of the larger data science software companies. His team had just closed a multimillion dollar deal and the CMO for the customer leaned in and asked, “OK, so now what do we do with this?” The moral of this story is get educated before you get in.
Knowing who to train and how much training is necessary is a function of involvement. Here’s where a key decision needs to be made. How involved will the current business be in the management of predictive analytics?
There are two major routes to success in predictive: become producers or be consumers.
Producers are businesses that make predictive analytics a core competency. This route requires the most education. There are a number of reasons to become a predictive producer:
The training regime needed to support becoming a producer can seem overwhelming but it doesn’t have to be. The business needs to train experts in managing and distributing predictive analytics. That doesn’t mean those people need to become experts on predictive methodologies or models. Minding that distinction is key to getting the right people in the organization the right kind of education.
Becoming a consumer requires less training intensive but is still skill intensive. Consumers will almost always outsource predictive analytics capabilities.
For outsourcing to work, the business must know how to manage external analytics teams. That means keeping them on task and on time as well as insuring the analytics products are producing accurate results.
Whether a producer or a consumer, the business is going to need talent. A recent estimate put the total number of data scientists in the world at around 150K to 200K. In my experience, about 1 in 10 data scientists have the skills to build predictive analytics tools so roughly 15K to 20K. That number will rise dramatically over the next year or two, but for businesses that need to stay competitive, that will come too late.
The business needs to get connected with predictive talent to succeed with predictive analytics. My most recent client was a startup. I chose them because they presented an opportunity to work with an amazing team to build an industry first predictive analytics solution from the ground up. That’s a common theme with predictive analytics talent. When we find a challenge that resonates with us, we’re all in. By revealing its challenges and connecting with predictive talent, the business can build the kind of interest that attracts top talent.
The paradigm of data science was a game changer and it still is for many late adopters. To move forward with predictive, the business needs to leave data science behind.
While that sounds like bad news, in reality it means some cost savings and improved capabilities for those already on board with data science. For late adopters, it’s a chance to get back in front.
Data science methodologies are associated with descriptive analytics, those that provide a rich picture of right now. Those methodologies do a great job of connecting multiple data points. Predictive analytics connects events; it’s a completely different approach and toolset. If you want a deep dive, check out my post on the difference between predictive and data science.
A lot of what was once a manual process for data scientists can now be automated. There are a number of high quality software suites available for managing data and producing high quality descriptive analytics. That means cost savings for the businesses already in data science as well as rapid ramp up for late adopters. Like I said, leaving data science behind isn’t bad news.
That lays the foundation for predictive capabilities. It’s time to start thinking differently about analytics. With data science, analytics tools were tactically significant. Predictive analytics tools are strategically significant. That means they can be used by new consumers and bring new capabilities to the existing analytics users in the business.
The largest impact and ROI from predictive analytics comes from the strategy planning process. There is an obvious and significant advantage to being able to see one step farther than competitors. Predictive analytics provide those capabilities. To leverage those capabilities right out of the gate means that the process has to be re-tooled away from dependency on descriptive analytics.
It’s a typical pitfall; applying old thinking to a new technology. This pitfall is one of the easiest ways to sap value from predictive capabilities. That means it’s time to stop thinking about analytics tactically and start viewing them as a strategic enabler.