Two days ago, we discussed in class how some companies (like IKEA or Shouldice Hospital) have figured out to turn smart outsourcing of work to customers into a win-win. Such strategy, also called co-production, can lead to higher customer satisfaction, involvement, ownership, and flexibility. (As he has proclaimed before on our Operations Room blog, my good colleague and blogger Marty often prefers self-service…)
But, like everything in life, there are trade-offs and we must seek a balance. Pushing co-production too far can be destructive; in The New Yorker, James Surowiecki describes the disasters at Circuit City and Home Depot. In retail, customers often are not fully informed about product selection, availability, location, and characteristics. Well-trained sales people can be of tremendous help, increasing sales (by reducing walk-outs due to frustration or phantom out-of-stocks), customer satisfaction (and hence future sales through repeat business), and employee morale (and reducing churn). Marty has blogged about our MIT colleague Zeynep Ton‘s study which found that four low-price retailers with much higher labor cost than their competitors also are more profitable. In another empirical study, Wharton colleagues Marshall Fisher, Krishnan, and Serguei Netessine quantified the increase in retail employee productivity:
“increasing associate payroll by $1 at a given store is associated with a sales lift of anywhere from $4 to $28, depending on the current level of payroll relative to store sales. The implication of this finding on retail performance is quite dramatic.”
I love facts.
But clearly, the sales lift will be concave in the staffing level as going beyond a certain optimal point will no further increase sales. The question then is: what is the optimal staffing level where marginal cost equals marginal revenue? Time and motion studies allow a reasonable understanding of the product-centric tasks (receiving and restocking) but the customer-facing service tasks are more nebulous and any quantification (which is needed for modeling) requires very detailed data-streams. But this is coming: INSEAD colleague Nils Rudi is embarking on very detailed data-gathering of customer-staff interaction patterns which we hope to report on soon…