This article is about scorecards, specifically performance scorecards. You cannot discuss scorecards without addressing the companion question: what is the difference between a scorecard and a dashboard? In general, I think of dashboards providing insight into current conditions, not unlike the dashboard of your car. A scorecard goes beyond a dashboard to relate the various measures displayed regarding strategy and goals. A car's dashboard does not really provide a goal - although that large number at the end of the speedometer does frequently tempt me. Well-constructed scorecards make strategy and goals apparent and relate current performance to expected results. A common element of scorecards is the key performance indicator (KPI).

A KPI typically has four elements: 1) a measurement, 2) a goal, 3) a trend and 4) a visual indicator. Let's say your company implements a quarterly customer satisfaction (CustSat)  survey. Your CustSat KPI may measure the percentage of customers who are satisfied or very satisfied. The goal for your part of the business may be 80 percent. Typically, KPI goals are multipart; 82 percent and above is considered good, 60 to 79 percent is bad and below 60 percent is ugly. This good, bad, ugly theme is common in KPIs indicated by up, down or level trending symbols. The indicator may be a traffic light. Your scorecard will have all of this information usually using the actual numbers for the current result and goal and glyphs or icons for the trend (often arrows) and indicator (traffic light in this case).

But I didn't really come here to talk about scorecards. I came to talk about how many KPIs a company should have.

I was giving a presentation during the SQL Server 2005 launch last year when an audience member asked me, "How many KPIs should a company have?" My immediate answer was, "Seven." People laughed; either because I answered so quickly or because seven is a funny number. I didn't pick seven to be silly. I picked it because seven is about the right number. As I said on stage, think about it ... You want something related to revenue, something related to costs or profits, something about inventory, something about customer satisfaction, something about employee health/morale, etc. I asked the person who raised the question how many KPIs his company had. In a rather sheepish voice, he said, "Three thousand."

We often think of a scorecard as the single page that we use to describe our business and its performance. In this customer's case, they have a set of encyclopedias that describe their world!

Personally, I think having 3,000 KPIs in a company means the system is broken. Maybe seven is not the right number, either. In the Balanced Scorecard (BSC) methodology, companies build scorecards for the financial, customer, operations and human resource areas. Using BSC, you could easily have seven KPIs for each area. But at some point, the KPIs become so fine-grained that they no longer unite people across the company in the pursuit of a common strategy. Managers cannot compare results across people or organizations, because the definition of success varies too much. The cost of managing and maintaining too many KPIs becomes a burden.

I tried to think about how I would come to have 3,000 KPIs, if I wanted that many.

First, I would allow lots of people to define KPIs. Each person in the company faces a unique set of challenges and sees the business through his or her own circumstances and experiences. Why not allow each responsible manager to define his or her own definition of success?

I believe if companies are going to profit from scorecards and KPIs, they should define their success metrics once, for the entire organization. Scorecarding is about the figurative single page. KPIs are the key elements of the scorecard that describe performance and achievement of goals.

Second, I would create KPIs at every interesting point in the business hierarchy. If I had three subsidiaries and cared about days-sales-outstanding (DSO), I would have a DSO KPI for each subsidiary. They might each have different definitions. Or, perhaps, we have consistent definitions of the DSO calculation, but each subsidiary has its own goal. Either way, I could have KPIs for each subsidiary. I'd have KPIs for each product. I'd have KPIs for each sales office and sales representative.

The way to cut down this version of KPI clutter is to make your KPIs multidimensional. Define them and calculate them inside your online analytical processing (OLAP) engine. A good OLAP engine can calculate KPIs using a multidimensional measure as the goal. That makes it easy to have goals that vary by subsidiary, by product, by sales office and by sale representative.

Third, I would let every business application calculate its own KPIs. It's true that the customer relationship management (CRM) application and your inventory management application will have largely nonoverlapping KPIs. But if your important business entities, such as customer, cut across multiple line-of-business applications, you will end up with multiple definitions of some KPIs. Perhaps worse, you will be working from nonintegrated data.

Using your OLAP engine not only gives you multidimensional KPIs, it also gives you consistency compared to each business application, portal and report doing separate KPI calculations. You'll have one definition and implementation of each KPI, and each will come from the same view of the business. That view of the business may be your data warehouse.

Fourth, I would think of a KPI for everything. We've all seen enthusiasm for new technology drive the ultimate overuser. I'm sure someone out there has a KPI about the number of KPIs. Perhaps my friend with the 3,000 KPIs ...

KPIs can be your entry point to business intelligence and business insight. When we first built SQL Server Reporting Services, we could not get enough reports, but in a short time frame we had more than we could handle. The phenomenon

of report overload is real. A well-designed scorecard with KPIs can help. When a user sees something interesting, he or she should be able to click on the KPI and see a report or a spreadsheet that describes the underlying situation. Now users are reading reports or spreadsheets that they want to read, and they probably know what they are looking for. And, to the relief of the IT department, users are consuming ready-made data sets, as opposed to traipsing around looking at everything.

My final point goes beyond the issue of having too many KPIs. The beauty of scorecards and KPIs are that they can drive accountability in your organization. The simple act of attaching employee names to KPIs is a sea change in understanding responsibility in the company. Imagine starting with a high-level scorecard. You see something interesting. You drill down.

As you do, the name of the responsible person changes. It feels a bit like rank-and-spank, but that is not my intent here. Yes, that will happen, but you will also find best practices and the people responsible for them. More importantly, people will watch their KPIs and hopefully take active responsibility for their performance. And that can only increase the performance of the whole organization.

Five Rules of Thumb for Key Performance Indicators

  • Define each KPI once.
  • Make KPIs multidimensional; allow the goals to vary by business entity.
  • Calculate KPIs centrally from your company's data warehouse so they all reflect the same view of the business.
  • Use KPIs as the entry point to business intelligence. Attach reports and spreadsheets to your KPIs. Use the smallest number of KPIs needed to show the true condition of the business, and to motivate and direct deeper exploration.
  • Hold people accountable.

Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access