BPM: How did you choose the metrics for these comparisons?

Erickson: A little bit of it was trial and error. Figuring out what metrics to use and then how to adjust for things like regional variances was not easy. You literally just have to try some of these things out. Take data into Excel and look at a couple regressions and see which one is giving you a higher correlation to performance.

Also, there is value to using consistent denominators. You don't want to be looking at things as a percentage of revenue for one department, a cost per unit for another, and a cost per resident-day for a third. The general services department handles facilities maintenance, the housekeeping, security, things like that. For those areas, we had to decide, what's a better metric: cost per resident-day or cost per unit? Looking at things on a per-resident-day basis naturally makes sense for us. We can improve the per-unit cost, but we're not getting revenue on empty units, so per-resident-day tends to be a better metric.

BPM: How does this tie in to budgeting?

Erickson: We have integrated performance evaluation, budgeting, and long-term planning. For reviewing performance, our metrics encourage managers to adjust spending according to changes in factors outside their control, like the number of residents. If you have 50 fewer residents in the month of June but measure performance by gross expense, a restaurant manager is going to spend just under what was budgeted. However, if you use a metric like total dining cost per resident, that manager is going to have to adjust expenses proportionally to prevent the metric from being worse than budget.

When it comes to forecasting, most people would think you first build your budget, then build your long-term plan. We do the opposite; a central team builds the long-term plans first. We run high-level linear regressions using historical financial data from our communities to generate net income metrics. In a sense, then, a long-term plan for a community is based entirely on how other communities performed at similar resident levels.

The long-term plan is then used as a tool for both setting budget targets and reviewing the budgets. When the final budget -- which is done at a much more granular level -- is rolled up and approved, if there are any significant discrepancies, we adjust the factors in the long-term plan accordingly so they are consistent. This extends our ability to identify problems well in advance. A budget might be 10 percent off the long-term plan because the regional factor wasn't high enough. When we make the adjustment to the plan, we can see how the community will operate once it is fully built out. If the margin is too low, it's a lot easier to raise monthly fees for people who have not yet moved in than to wait and raise fees higher than COLA for everyone.

BPM: How do you ensure that line-of-business managers aren't cutting costs so much that their efficiency reduces revenues?

Erickson: All these cost targets are balanced with resident-satisfaction statistics. Some campuses that have extremely efficient staffing ratios are not able to meet their resident-satisfaction numbers. We're always wanting to keep those in balance. The executive director of a community is evaluated pretty evenly based on financial performance, employee satisfaction, and resident satisfaction. He's being evaluated on the percentage of people saying yes to satisfaction questions like "Would you recommend an Erickson campus?" All the dining directors are measured off of the dining satisfaction performance scale, as well as their financial metrics.

BPM: How have line-of-business managers reacted to being judged in this way?

Erickson: I would say pretty positively, especially the top performers. If you want to drive to better performance, you have to be able to reward top performers, and you can't reward them unless you can identify them. This system allows us to identify the people who perform above average, and I think people have accepted that comparisons are going to be part of it.

We don't just have one totally objective model that's crunching numbers and saying, "Here's your grade, and here's your rank." There are obviously regional things that factor in. You have snow-removal costs up in Boston that you don't have down in Texas. You can't do it completely mathematically. If all you're measuring people on is their ability to meet a budget, you're going to have people sandbagging and overestimating what their expenses are going to be. But if you have good targets in place, then you're not going to have that going on.

Before, there was a lot of politicking in the process, with people trying to build cushion into their budgets. Today if someone tries to do that, they essentially are saying, "I'm going to be 10 percent worse than my peers at this resident level -- and this is why." Nobody wants to say that. Having justifiable targets drives more accurate and more aggressive budgets. The top performers like it, and those are really the people who you want to cater to with a system like this.

BPM: What results have you achieved through this approach?

Erickson: It's interesting: When you plot out these data points, you can see that the newer communities have been performing better than the average from prior years. The last two communities we opened had more efficient staffing than the average four or five before that. And the communities that are finishing up today are more efficient in that final stage than they were a couple years ago.

BPM: How did you get to this point?

Erickson: This was a long, iterative process. Budgeting was why we looked at Hyperion in the first place. We used the implementation as an opportunity to reengineer our budgeting process. You can take your existing budgeting process and mimic that with a BPM solution, but you're not going to get as much value. Change of this magnitude can be disruptive, so a lot of credit for our success is due to our CFO, Jeff Jacobson, for his role in setting the direction and communicating the value of this initiative to the leadership team.