We use cookies. You have options. Cookies help us keep the site running smoothly and inform some of our advertising, but if you’d like to make adjustments, you can visit our Cookie Notice page for more information.
We’d like to use cookies on your device. Cookies help us keep the site running smoothly and inform some of our advertising, but how we use them is entirely up to you. Accept our recommended settings or customise them to your wishes.

Data Integration – Thinking Beyond CDI to Predict the Future

When most people think of the term data integration, they think of CDI, or Customer Data Integration. This is basically the process of integration data at a person level through various pieces of information. In a previous blog, I talked about the idea that CDI has primarily been based on name and address for the past 30 years in direct marketing, but how that is shifting to a many-to-many matching issue created through all the digital channels.

In this blog, I’d like to explore the other ways of integrating data. What is data integration anyway?  It’s basically a merge two different data sources and the fundamental basis of any merge is a primary key. Well, we’ve discussed the idea of matching person-level information, but what about other types of data integration?

Other than at a person level, I think there are two simple data integration concepts that are highly under utilized. These are location and time. Consider time. The one thing about time is that it is a global standard. Everyone follows the same calendar – seconds, hours, days, weeks, months and years. This creates an extremely easy method of matching and merging data. 

Basically any data with an associated timestamp can be transposed by time. The types of data are endless here – purchase transactions, site visits, store visits, social likes, search teams, media impressions & GRPs, weather, stocks, and various economic indicators. And the most logical application is using this time-series data to build a better forecast (of anything).

I think improved forecasting is a huge opportunity in analytics today. Think about any organization and the better decisions they would make if they could know how many times the phone would ring, or how many shoppers they’d get in store #252 or what inventory levels would look like. Every company has the need for a improved forecast, but I’d argue the sophistication and rigor put into those forecasts are probably fairly rudimentary. Sure some organizations have this down cold knowing how many widgets they will sell in a particular store, but do they apply that same rigor to other areas in the business?

This is where more and better data in building a forecasting engine comes in. How would you ever know if buzz on Facebook is predictive in determining the number of Happy Meals that McDonald’s sells in a given area if we don’t analyze that data in a meaningful fashion. I think the opportunity to organize any data that can be summarized in the dimension of time is a missed opportunity today in business analytics. And furthermore, the need and application of improved forecasting is a huge one that typically goes unnoticed, mainly due to lack of awareness of a better way to do things.

In a future post, I’ll discuss the same idea, but with location data. 

Join the Discussion