Market forecasting tools can tell you how business your company is winning and your aggregate market share. But what if you want to know how much of each client’s business you’re winning?
That’s the question we at IBM set out to answer several years ago. And it’s a case study that we analyzed as part of the Center for Applied Insights study, “Marketing Science: From Descriptive to Prescriptive.” We thought that the data issues that IBM had struggled with might shed light on how other companies can use marketing science in their businesses.
In this particular instance, IBM had to figure out how we could measure our share of each company’s tech budget.
We started out on square one, figuring out which organizations we could serve. This meant identifying every company – and every one of its subsidiaries – in all the markets we operate in. Pulling together this information was a big job, but it turned out that we faced an even bigger hurdle (you can probably guess it).
We had lots of customer data: 10 terabytes, to be precise. And all of that data was sitting in different databases throughout the company, each built by individual business units over the years. And the data definitions they used varied widely. The top 164 customers alone had more than 600,000 unique identification codes. To get a coherent view of IBM’s business at the client level, we would have to reconcile millions of customer identifiers before cleaning up and consolidating the data.
A few months later, we finally had a structured database. The last step was to work out how much each customer was spending on IT.
We did this by using another forecasting tool we had developed that estimates the size of the overall market along with mathematical algorithms to assign a share of that expenditure to every company, whether they were customers or not.
It was a huge effort, and one we've refined many times since, but one that’s now paying handsome dividends. With this view, our sales and marketing teams can detect buying patterns and target holes in IBM’s market coverage. And by integrating the insights the view provides with other customer data, we can identify new opportunities. We can also explore why specific customers buy (or don’t buy) particular solutions, why some don’t spend very much with us, and what we can do to serve individual customers better.
That’s lets us anticipate and respond to our customers’ needs and allocate our resources more effectively. And we think it’s an example that demonstrates just how powerful marketing science can be within a company.
The IBM CAI blog has moved - please bookmark and follow the new blog at http://www.ibmcai.com