Oracle Retail Blog helps retailers stay current on customer successes, hot retail trends and industry best practices.

How Decision Science Helps Snag the Stanley Cup and Shape Retail KPIs

Mark Lawrence
Solutions Director, Analytics and Enterprise

Apple, Beauty and Biscuit. No these are not the names of celebrity children. I am talking about hockey as a metaphor for KPIs. With the NHL playoffs right around the corner, my thoughts naturally center on KPIs and analytics. So lets get our Mitts and Twigs ready and dive into how these things are related.

For example, when down by 1 or more points, it is common for coaches to replace the goaltender, known as pulling the goalie, with an offensive player in the final 1-2 minutes of the game as a last-ditch effort to at least bring the game to a tie and force overtime. Of course, exposing the goal this way involves some risk-reward analysis, but could result in a Celly. If your sole objective is to win the game, exactly when is this best done?

Several recent studies employing decision science have revealed that the optimal time to pull the goalie when down by 1 point is actually with about 5 minutes left in the game, and with about 11 minutes left when down by 2 points. But, while the science at least appears to be influencing coaches to pull goalies slightly earlier now, it’s never by nearly that much. Why is that? Don’t they want to score a Gino and win?

Well, they do, but apparently winning isn’t the only objective. Another objective might be optimizing fan sentiment, which would be harmed by a higher point-margin loss resulting from pulling the goalie earlier, especially for fans that paid a small fortune to see a live Gongshow.

Recall in a previous post, New KPIs of Retail, the recommendation to tie KPIs to objectives and objectives to goals, for context in how to interpret them to inform critical decisions. If one of those two objectives above (winning and optimizing fan sentiment) weren’t considered, the decision on when to pull the goalie would be very different. Goals, objectives and KPIs are, wait for it…a trifecta! 

Of course, Retail KPIs are no different. Without the context of a company’s goals and objectives, KPIs might not be properly applied to strategic decision-making. Gleaning insights from KPIs is the easy part – putting those insights into proper context to inform decisions is the more challenging part.

Oracle Retail Insights enables the mapping of KPIs to goals and objectives for a true balanced scorecard view, instead of a partial face wash view.

Inspired by the when-to-pull-the-goalie example, I thought I’d share my top 10 list of commonly-referenced KPIs that warrant heavy consideration of context, from such a balanced scorecard perspective or otherwise, before drawing any conclusions and calling in the Goons.

Top 10 Common Retail KPIs

  1. Conversion Rate % - The percentage of shoppers that made a purchase while in the store. It is derived by (Trx Count / Store Traffic * 100). 
    How should 'Buy Online Pickup In Store' transactions be considered?
  2. Comp Gross Sales Amt - Value of units sold in comparable stores. 
    As assortments become increasingly localized, how to determine a store as truly comparable is becoming more of a challenge.
  3. Average Trx Amt - Amount on a transaction. 
    Not a particularly nuanced KPI when 100% of transactions are in stores, but the nature of transactions can vary wildly with more complex combinations of retail channels.
  4. Avg Market Basket Size - The average number of units purchased by customers in each purchase at a trade area. 
    Trade area definitions, and changes to those definitions over time, can confuse interpretation especially in an increasingly omnichannel marketplace.
  5. RFM Categories - The score indicating the combined Recency, frequency and monetary value of a customer. 
    In aggregate, at least to inform your strategic customer segmentations, I’d argue that this should be a standard KPI, albeit a lagging one. RFM is typically in this order because of presumed importance, but is recency truly the most important for ALL retailers?
  6. Gross Profit - Difference between sales revenue and the cost of units sold. It indicates the retailer’s ability to mark-up merchandise for sale. 
    As we go from forest to trees to individual leaves, interpreting this super-critical KPI requires special care.  What if a less profitable subclass is often an anchor to a more profitable subclass in a basket?
  7. Reg Out of Stock % - Percentage of regularly priced unique items that were out of stock as of the last batch run. 
    Stock-outs are bad no matter how you look at them, but as a percentage, and further sliced across dimensions of product and location, increases or decreases require interpretation.  And if minimizing stock-outs are a concern, consider Store Inventory Operations Cloud Services.
  8. Sell Through Retail - Ratio of the gross sales amount as a fraction of owned inventory for a given time period. 
    What if at least some of your inventory is consignment?
  9. SL Shrink Retail - Retail value of inventory lost through means other than a sale. This is the difference between actual physical inventory counts and the amount of inventory reflected in the stock ledger. 
    To what extent does this indicate theft? Process failures? Spoilage?
  10. Customer Lifetime Value – OK, while very apropos to this list, I’ve included this one to lure you to read New KPIs of Retail Part II in case you haven’t already.

I selected these from among the 13,984+ packaged measures, metrics and KPIs of Oracle Retail Insights Cloud Service Suite v18.  Note that this count includes various metric grains (WTD, MTD, YTD, etc.) as well as “As-is” and “As-was” views to account for item reclassification's and store relocation's. 

Watch Retail Home Video

Retailers don't need to worry about sitting in the Sin-bin, Oracle helps ensure that these and other metrics are properly cast to end-users in a manner that provides perspective, and context, beyond enabling balanced scorecard views. The packaged names and many of the derivations are configurable, and of course can be surfaced to Oracle Retail Home to:

  • CONNECT users to the information that they need
  • ANALYZE in Oracle Business Intelligence
  • EXPERIMENT in Oracle Retail Science Platform
  • EXPLORE in Oracle Data Visualization or exposed to other solutions via RESTful services
  • ACT in an informed and timely manner


There is no need to call in the Stripes. This is truly and fully a model-once-deploy-anywhere approach to retail analytics and decision science. It is not a schema-on-read Data Lake (where a user is entrusted to properly interpret raw data), but rather a highly-governed schema-on-write Data Warehouse, with the ability to augment with a Data Lake where specific use cases warrant.

And, similar to how NHL coaches are considering both when science tells them to pull their goalies along with their own intuition, Oracle Retail Insights Cloud Service Suite enables Adaptive Intelligence (yes, yet another of my favorite terms) which enables a balance between artificial intelligence and human judgment through a blend of extreme configurability, as well as a flexible array of federated tools and pervasive user experiences.

I'm looking forward to the playoff's and to see who is going to take home the Cup this year. Go Penguins!  

Request a Demo

Be the first to comment

Comments ( 0 )
Please enter your name.Please provide a valid email address.Please enter a comment.CAPTCHA challenge response provided was incorrect. Please try again.