Article

Ten best practices for implementing and fine-tuning retail metrics

Dashboards provide a clear picture of business operations, provided the metrics behind them are properly designed.

September 21, 2010

For busy retail executives and store managers, metrics and dashboards provide a clearer picture into their business operations, allowing them to manage and grow an individual store and an entire chain with confidence. Successful practitioners have learned how to leverage these tools to transform their businesses from me-too organizations to market leaders.

But what if the metrics behind the dashboards aren't right? Poorly designed metrics can do serious damage to a business. Managers responsible for customer service and satisfaction can be distracted from what really matters: improving the customer experience.

In recent years, standard metrics such as the Net Promoter Score (NPS) and the American Customer Satisfaction Index (ACSI) have been wheeled out as cure-all customer satisfaction metrics. While NPS and ACSI are useful in certain situations, many companies have become over-reliant on these relatively one-dimensional metrics. There’s also no shortage of companies suffering from the proliferation of metrics. Managers are given reports with metrics for everything ranging from average checkout time to number of complaints registered to average availability of produce.

Here are some best practices to achieve quality metrics that customer-focused managers can rely on:

1. Tie metrics to corporate business strategy.Too often, organizations show common, "cookie cutter" metrics for every project. A clear definition of the business strategy should precede the development of each set of metrics. For example, if a retailer’s strategy is increasing its average transaction size, its metrics should reflect customer satisfaction as well as purchasing behavior.

2. Metrics should be reliable. Reliability of metrics depends on the reliability of underlying data and the composition of the metric. Preferably, the reliability should be scientifically tested. As an example, if a store wants to increase its average transactions size, its customer satisfaction metrics should be analyzed against transactional data to ensure a strong correlation. If customers with high customer satisfaction metric scores don’t have bigger transactions than their “unsatisfied” counterparts, it’s time to reevaluate the metric.

3. Metrics should be specific. Nearly every organization wants to improve its customer satisfaction and loyalty and relying on a single customer satisfaction metric (NPS, for example) can be tempting. However, metrics should be specific to an organization and link to its strategy and core values. For instance, if the retailer sets a goal to reduce customer complaints by 25 percent in one year, the company needs to track a complaint and resolution metric on a weekly basis.

4. Metrics should be hierarchical.A very limited number of metrics (one or two), should be used at the top level. Each key metric should have sub-metrics, like in a family tree. In this case, single customer satisfaction metrics, such as NPS, are useful. Think of it as the tip of the pyramid. To be helpful, managers need to be able to see the whole pyramid. A low NPS score should be supported with layers of more specific metrics, e.g., satisfaction with pricing, store layout, speed of checkout, associate knowledge, etc., that inform specific responses.  

story continues below... advertisement
		The Best Retail Customer Experiences 2010 Even with a slow economic recovery, people are still spending money in the retail environment. RetailCustomerExperience.com surveyed more than 1,000 customers to discover the most popular retailers in the country. Find out which retailers come out on top in customer service, aesthetics, online experience and inventory.   Order your copy today!

5. Metrics should reinforce competitive advantages. Metrics should reflect an organization's differentiation and competitive advantage. If an organization's competitive advantage is its ability to introduce new and innovative products, then revenue growth from new products introduced in the last three years is a more important metric than overall revenue growth.

6. Value rank the metrics. How do you decide what to show when your dashboard has 20 charts and tables and each is important? Customer Satisfaction dashboards often suffer from this problem. They have so many different measurements, that a manager doesn’t know where to focus or picks something different each month and gets nowhere. Value ranking is one way to decide what to show where. Ask the likely users of the dashboard to assign values on a scale of 1 to 10 for all the dashboard elements. An even better way is to assess which metrics actually drive business performance through regression analysis. Keep the highly correlated metrics and throw out the rest (or put them in a different report).

7. Use weighted metrics. Weighted metrics compound underlying data by assigning different levels of importance to various data elements. The weights should be scientifically determined. As an example, many retailers have moved beyond asking their customers directly “How satisfied are you?” and are measuring their customers’ satisfaction through the customer experience. By analyzing how customers answer the questions about their experience against their satisfaction—measured by an additional survey question or by actual purchase behavior—a savvy company can determine exactly what matters to their customers and, ultimately, how to create an accurately weighted customer satisfaction metric.

8. Establish a common value/scale. All metrics should be normalized to a common value. The best such value is 100 with a scale of 0 to 100. By having a common scale, users do not need mathematical calisthenics to compare one metric with another.

9. Simplify dashboards and reports. Pretty charts and a lot of numbers do not constitute a useful dashboard. Good dashboards are multilevel, with the top level providing the most useful information, with as few metrics and as few plots as possible. Death by data and charts is a common disease of most dashboards.

10. Employ real-time metrics.Metrics and dashboards should be available on demand, and the metrics should reflect real-time or near real-time data. Web portals are a convenient way to deliver the data, allowing structured access to data based on permission rights. In the retail world, for instance, trends change from week-to-week and even day-to-day. Think about the period between Black Friday and Christmas Eve. Getting data for this time period in the following year won't help a store influence sales or customer satisfaction in the moment.

Metrics can help guide strategic retailer decisions that impact customer experience and ultimately sales but it is easy to get buried in data. Retailers must have a strategy behind the metrics — how they measure and share the insights — for the greatest reward.

Dr. Pawan Singh is CEO and chief science officer for business intelligence firm PeriscopeIQ. (Photo by Jorge Franganillo.)