3 steps to get your company’s data program back on track

If your company is having trouble seeing concrete results from your data program, it might be time to start over.

Here’s how to do it — and what senior executives need to do personally in the process.

Also read: How to prepare your company for incidents with artificial intelligence

For a new beginning, companies must follow three fronts in parallel:

1. Attacking data quality

Businesses are plagued by a wide range of common data quality issues.

The key is to minimize “hidden data factories” by reducing errors at the points of creation. And the key to this is for Sales to clarify what they need, discuss these needs with Marketing and for Marketing to eliminate the root causes of errors, one by one. The benefits are easy to see and appear quickly.

2. Starting the data science journey with simpler techniques

Data science is ideal for tackling all types of problems that companies face. In order of increasing complexity, some methods and the problems they help solve include:

— Statistical process control and improvement are unbeatable for reducing costs, cycle time and uncertainty in business processes.

— Experimental design (including A/B testing) helps improve product quality.

— Statistical modeling helps companies understand customer behavior, predict next month’s sales, decide where to locate a new store, and estimate the fair price to pay for an acquisition (among other examples).

— Machine learning/artificial intelligence, used to analyze large data sets, reveal previously unknown relationships, improve decision making, and automate many types of work.

To get started, leaders should start at the top of the list above, select a few key business processes (again, led by open-minded people), ask everyone to put together a small team, learn about statistical control, and then do the work necessary to get their processes under control.

Learn along the way, but to be clear, this exercise isn’t just about gaining confidence in data science. Expect to see real business benefits in no time.

3. Putting data to work and creating competitive advantage

While AI has garnered most of the attention over the past year, there are many other ways to generate business benefits from data. Four great examples include:

— Make products and services more valuable by incorporating more or better data. This can be done with virtually any product. My favorite example is the Coors beer can — the mountains turn blue when the beer is cold enough to drink.

— Package and sell data products.

— Create sustained advantage by exploiting proprietary data — that is, data that no one else has. A good example is S&P’s Cusip, which uniquely identifies financial products.

— Bring more and better data to decisions, fulfilling the promise of “data-driven decision making”.

Over the long term, companies will need to choose one or two of these options that best suit them, based on their skill sets, the data they have or can obtain, and the markets in which they compete.

Incorporating data is almost always a good starting point, in part because the work is straightforward: Ask some of your best marketers to document their understanding of what customers really want and need.

Then conduct a simple “gap analysis” to determine what they would like that you are not delivering. Try to fill these gaps with data.