KPI: Leveraging corporate data with key performance indicators

Employees who monitor KPIs now receive a daily digest e-mail with the relevant metrics for their roles. "The e-mail arrives early each day, and our staff can scan through the  metrics." continues Smith. "If there are issues they need to explore further, they can do additional research through Metric Insights or go to our query tool." Metric Insights also offers a dashboard, but RealPlayer staff members prefer the e-mail. The company is also considering an alert system in which users would get a message if a KPI threshold was crossed.

Change notification

The key performance indicators (KPIs) that RealPlayer focuses on are customer acquisition and monetization, but with the advent of Metric Insights, the company is expanding its scope. "We are now starting to look more at the metrics associated with the impact of new features on user retention," says Smith. "Also, since we can connect to Google BigQuery, we are able to look at patterns of usage across the entire spectrum of our user group."

Metric Insights has made the monitoring of KPIs more efficient and consistent. "In the past, business owners would query our database, find out their KPIs and then send them out in a spreadsheet," says Smith. "Now, everyone gets the message first thing in the morning and they know what they need to address."

The approach used by Metric Insights is to leave the data in its existing repository and to pull summary data for the individual KPIs into a KPI warehouse. It is designed around notification so that users are alerted when the KPIs change. Steve Mock, COO of Metric Insights, disputes the metrics of the traditional dashboard. "If you log in and look and there is no change, your time is wasted," he says. "With our approach, you can see in 60 seconds all the metrics that are important to you."

The model also has a collaborative dimension. Users can make comments that are shared by others who access the same KPIs, according to Mock. Annotations may also be made automatically by the system, which indicates correlations to events that may show a causal relationship. Users can also customize their KPIs by selecting some as favorites and can put into other applications the visualization charts that are refreshed at a specific frequency.

What makes a good KPI?

The best choice of a business metric is usually a ratio that reflects a fundamental issue in the health of a company, according to Alistair Croll, an entrepreneur and co-author of Lean Analytics. "The number of guests per day in a restaurant or the revenue per person are more meaningful metrics than the total number of people who have come to the restaurant over the years," Croll says. "These measures can be compared to fundamental business assumptions such as ‘We will have 50 guests per day.' or ‘Each of our diners will spend at least $20.'"

Croll advocates focusing on a single metric at any given point in time. "A business should be aligned to get the one most important variable to move to where it should be," he explains. The value of having one variable is that it allows for a controlled experiment. "Many companies underestimate the importance of experimentation," he says. "If you change too many things at once, you don't know which one caused the change in performance." Companies should develop an ongoing process during which they optimize one factor at a time, and then move on.

Total revenue is a metric often shown on dashboards and, naturally, all companies monitor it, but Croll does not consider revenue to be a good KPI. "Revenue is an output that depends on other activities, such as the rate of production or the performance of a salesperson," he says. "A measure such as revenue per salesperson is a better metric." Moreover, if employees have awareness of the impact of their actions on a particular metric, they can modify their behavior to achieve the desired outcome.

The two most common mistakes in using KPIs are lack of focus and failure to connect the KPIs to the business model, according to Croll. "So much data is available, it's easy to throw together an infographic to hide all manner of sins," he says. "Companies need to be willing to try different things and determine what they would deem a ‘success.' These are cultural and organizational issues that need to be tackled if a company is to become data-driven."