News Stay informed about the latest enterprise technology news and product updates.

5 ways to think like a data scientist without becoming one

CIO Symmetry occasionally posts expert advice we think will be of use to our small business readers. Bill Stripp, director of IT for NxtTeam, a managed service provider specializing in business intelligence for small to mid-size businesses (SMBs), believes data analytics is not just for big companies. He offers five steps SMBs can take to become more “data-driven” and boost profitability.

A Bill Stripprecent survey by the Economist revealed 59% of respondents who described their organizations as data-driven reported greater profitability than competitors, compared to those polled who said their companies are not data-driven.

The findings are not surprising. Hidden within the data flowing through your company are insights into every function — from inventory to sales to finance — waiting to be translated by into competitive advantage.

But how do companies unlock those insights? If you’re a small or mid-size business (SMB), chances are this info-opportunity is a struggle. Most business intelligence (BI) and data analytics tools are designed for large organizations — too costly, complex and staff-intensive for the limited resources of SMBs.

But the landscape is changing. With the advent of cloud-based BI applications and services, SMBs of all sizes and types now can act like big corporations, with a team of data scientists plying the requisite coding and analytical skills. Cloud delivery means the high initial cost of entry and maintenance can be reduced to a smaller manageable monthly fee for service.

BI as a service brings the mining and analytics capabilities of a data scientist to employees through a web-based portal. Members of any department can gain access to relevant business data specific to their role and presented in easy-to-digest formats. Business teams can enjoy the benefits of self-service BI without depending on an expensive multi-year implementation.

What should they do with that information? Here are five ways a data scientist would approach the data:

Think Q&A

Many BI projects fail because some companies approach information management like playing Jeopardy. They see analytics as answers and start looking for questions to match. That strategy, of course, is backward. Data scientists determine the right questions before they start asking data for answers. The best questions to pose first relate to an organization’s goals and strategies. Other equally important questions don’t relate to the data itself, but the processes and procedures that organizations use to collect and store their data, such as: Where do we find the right data to query? And are we collecting all the right data for analysis?

For instance, a manufacturer faced with a sluggish production cycle might pore over reams of reports about units per hour coming off the assembly line. But a data scientist would explore other factors, too, such as the number of raw materials involved in production and where those suppliers are located. Now, the manufacturer can ask: Do we know how long each supplier takes to deliver raw materials? And do we know what impact that timing has on units per hour?

Logistics managers could then consolidate sourcing by tapping into geographic data to find the optimal mix of speedy suppliers.

Think outside the lines

If the manufacturer in the first example limits analysis to the information generated by its own organization, the company may never accelerate production. Data scientists know key factors sometimes reside outside the literal and virtual boundaries of the business, especially in the era of big data.

The spectrum of data sources available to today’s SMBs ranges from internal streams — such as point-of-sale stations and warehouse intake systems —  to external torrents — such as email, websites and even social media. Data scientists consider data destinations, too, which likely are multiple, disparate databases, files and other systems.

A good model crosses virtual boundaries and helps identify relationships between different types of data. Cause-and-effect patterns emerge, explaining why, for example, a particular product sells better at certain times of year or in certain regions.

Think apples to apples

Like the variety of a fruit basket, a typical organization’s data is a collection of diverse files, databases and repositories. And just as apples, oranges and bananas have different shapes and textures, not all data is configured according to the same rules or processes.

Data scientists understand that extracting meaningful insights from data is a bit like making a tasty fruit salad: At some point, each category of fruit must be reconciled to determine the freshest pieces. So, too, must data be reconciled for analytics.

Just as tastes in fruit salad vary, data should be reconciled to fit the particular needs of a business. The goal is finding the best data salad to satisfy the information appetites of different staff in different departments – and make it the same way with the same quality every time.

Think blueprints before tools

In their pursuit of data insights, some businesses think like carpenters, electricians or plumbers. When evaluating BI technology, they gravitate to tools most familiar to them. So, sometimes they equip themselves with hammers when they have a bunch of screws and pipes to assemble.

Data scientists think at a higher level — like architects drafting blueprints. They consult with different departments to learn what they need from their data. Building your best analytics will take hammers, screwdrivers and wrenches to manage the nails, screws and pipes. But you won’t know how many and in what combination unless you understand the blueprint before buying the tools and hardware.

 Think in specifics

Data scientists know the objectives of BI shouldn’t be vague. Without specific metrics and key performance indicators (KPIs) aligned to specific company goals there’s no means of objectively evaluating success when implementing corporate strategy. Measurements must be definitive and concrete to indicate progress. For example, initiatives to expand geographically should point out destinations; campaigns to boost profits should define margins; a move to cut costs should enumerate savings; programs to boost productivity should quantify gains, and so on.

Becoming a “data-driven” company is a matter of mindset. But your team members don’t have to become data scientists to think like them. With subscription-based cloud services, SMBs of all shapes, sizes and industries can start to develop the same level of competitive advantage enjoyed by large enterprises – without the same cost, complexity and commitment of staff.

 About the author:

Bill Stripp is Director of IT for NxtTeam, a managed service provider specializing in business intelligence for small to mid-size businesses. During his decades-long career in technology, he has held technical, analytical and training positions specializing in enterprise software for Waste Management, Inc., Fujitsu Consulting and Greenbrier & Russel.