For organizations with complicated corporate or budget structures, keeping master data accurate -- and so keeping budgets and financial reports accurate -- can be surprisingly time-consuming. McKesson Provider Technologies, a subsidiary of health-care giant McKesson, has used master data management software to reduce frustration and free up employee time for more valuable pursuits.

BPM Magazine:What software does McKesson Provider Technologies use for budgeting, planning, and forecasting?

Heather Capell:Our general ledger's SAP. Our forecasting is done in TM1 (OLAP). Our budgeting is done with arcplan, interfacing into TM1. And we manage all of the master data that we import into the TM1 database with +EDM -- cost centers, profit centers, how they roll up into various hierarchies.

BPM:Why would you need to manage that information outside of your budgeting and forecasting applications?

Capell:We have three or four hierarchies of the same set of profit and cost centers. One rolls up by business unit; one rolls up by who manages or owns the profit and cost centers. One's by revenue stream -- so all our software-related profit centers roll up one way, all our services roll up another way, all our hardware sales roll up another way. And then we have a legal-entity rollup, just by company code. So we have attributes associated with the profit and cost centers like which company code they're in; which cost center; which class (is it R&D, is it G&A, is it marketing, sales?); and which revenue stream it's in. Using master data management software helps us manage the hierarchies, the building of the basic master data and then the attributes associated with them.

We set up all the profit and cost centers, the hierarchies, in +EDM and then we import that into TM1. And we have little tweaks and changes in our business units pretty much throughout the year. It's not just once a year during budget time or at the beginning of the fiscal year. Stuff is constantly shifting around, and +EDM helps our forecasts reflect that easily with minimal heartache.

BPM:What was master data management like before you had software dedicated to it?

Capell:Since we have so many hierarchies, it wasn't just setting up a cost or profit center and dropping it in its place. Each cost and profit center is mapped several different ways, and the way I used to build the hierarchies was very manual. I had a spreadsheet that had three different sections with the rollups. It basically just helped build the hierarchy, so not the actual data itself, but where the data went and how to handle it. That spreadsheet was used to handle all aspects of our reporting. Actuals, forecasts, budget would all depend on those hierarchies to organize the data. I would build the cost and profit center and business unit hierarchies in the spreadsheet in a test environment, copy that file to the server, and then bring the server back up and it would recalculate where it needed to.

So it was very time-consuming to make any type of big change. I'd have to go find all the places in the spreadsheet where a cost center was referenced, all the places where a rollup was referenced. And if I was changing the structure, the consolidation points, I'd have to go find everything on up the food chain. Say we wanted to move one business unit from the major business unit grouping it was in over to another one. So to move the business unit, I might want to recode everything in that consolidation point. The old code might have been PCA and now it's going to be RCB. I'd have to go through every section of that rollup -- and there might have been 40 or 50 consolidation points referenced in the spreadsheet. I'd have to go through and find them and make sure I recoded them properly, and if those consolidations were involved in multiple hierarchies, go to the other places on the spreadsheet or in the database to do that. I couldn't rename it one place and have it instantly be cross-referenced and updated.

BPM:Did this manual process ever lead to problems with developing your forecasts?

Capell:Sure. If I didn't make that consolidation point connect to the next point above it, I'd have something stranded -- mapped to a part of the structure that no longer connected to the rest of the hierarchy. And I could double-map it very easily if I added it at a different point in the spreadsheet. It was very easy to make errors of omission and double-counting. So we'd miss or double up on revenue or an expense during the forecast development process. Those two things happened frequently.

BPM:How would you discover an error in your master data?

Capell:Until you actually got the forecast data or budget data or actuals data into a new cost center or profit center, it wasn't really visible that there was a mistake, that it had been double-counted or that it had been orphaned. It would be right in the middle of crunch time, during forecasting, when they'd realize that a new cost center was mapped twice, that the system was double-counting. Or that a new cost center wasn't rolling up at all. Right during a budget deadline, I would have to spend time fixing that.