For the past couple of decades, retailing in the U.S. has -- with some notable exceptions -- been a vast experiment in minimizing labor costs.

At the 2009 annual convention of the National Retail Federation, though, Charles DeWitt noticed the beginnings of a shift. "Retailers started coming up to me and saying, 'We can't get any more out of this cost stone,'" recounted DeWitt, vice president of business development at workforce-management-software maker Kronos.

Since then, this change in attitude has become the stuff of business headlines. Most notably, Wal-Mart, the retailer that set the cost-cutting tone in the 1990s, has been raising wages and spending more on training. There's surely a cyclical element at work here -- as the unemployment rate drops, it's harder for retailers to find workers. There's also a political element -- bad press and minimum-wage campaigns must have some effect on corporate behavior.

But the really intriguing possibility is that retailers, in their technology-driven rush to optimize operations during the past two decades ("rocket science retailing," one Wharton School operations expert dubbed it) were actually failing to optimize labor. Their systems measured it only as a cost, and didn't track the impact of low wages, part-time work and unpredictable work schedules on sales and profits. Now some retailers are trying to fix that.

One big set of targets are the scheduling systems that have allowed retailers to ever-more-closely match staffing to customer traffic, but in the process wrought havoc with many workers' lives by making their schedules so unpredictable. Jodi Kantor gave a face to this last year with a compelling New York Times account of the chaotic life of a single-mom Starbucks barista.

Kronos supplies Starbucks' scheduling software, and DeWitt was quoted in the Times article describing its workings as "like magic." So it was a little surprising to see him on stage last week at O'Reilly Media's Next:Economy conference, nodding pleasantly and occasionally chiming in as a Starbucks barista, a labor activist and a journalist described the horrors inflicted by scheduling software.

When I told him afterward that I was surprised he wasn't more defensive, DeWitt said, "I'm more of a math guy, an optimization guy. This is a parameter to be optimized." It's also a business opportunity. "We are in early-stage investigations with very big customers," DeWitt went on. "The plan is to go in and suck all these things out of the database and work with them to customize metrics."

The idea is to figure out how dynamic scheduling and other labor practices affect metrics such as absenteeism, turnover and sales. Right now a lot of retailers just don't know. Carrie Gleason, director of the Fair Workweek Initiative at the Center for Popular Democracy and the labor activist who shared the stage with DeWitt, recalled a conversation she had with an executive at a big retailer at last year's National Retail Federation convention. "I said, 'These schedules cost you in terms of turnover.' She said, 'I’m in operations. That’s HR.'"

That's not true everywhere. Here's Stuart B. Burgdoerfer, chief financial officer of L Brands, the retailer that includes the Victoria's Secret and Bath & Body Works chains, speaking at the company'sannual investor day this month:

As we looked at the data, we just had too many people working too few hours per week. And the trouble with that or the opportunity with that is how well can they really know your business, how invested are they in us, or we in them, if they're only working a few hours per week and their turnover rate is very high?

And so we see the opportunity to have a more knowledgeable, more engaged, more effective and productive associate. When she's working, typically she is working more hours per week. So that's the opportunity. And we think it's a significant one. Really do.

Recent academic work backs this up, to a point. Researchers such as University of Chicago social psychologists Susan Lambert and Julia Henly and Pennsylvania State University labor economist Lonnie Golden have been documenting the extent and social costs of irregular scheduling. Meanwhile, operations experts at business schools have been trying to identify labor practices that maximize sales and profits.

The best known of these is probably the "good jobs strategy" outlined by Zeynep Ton of the Massachusetts Institute of Technology, first in a2012 Harvard Business Review article and then in a 2014 book. Ton studied low-cost, high-wage retailers such as Costco, Trader Joe's, Oklahoma-based convenience-store chain QuikTrip and Spanish supermarket chain Mercadona and concluded that they operated in a virtuous cycle in which highly trained, autonomous, full-time employees working with a limited selection of products drove high performance.

There's a tendency, upon hearing accounts such as Ton's (she also spoke at the Next:Economy conference), to wonder why every retailer doesn’t do that. One reason is that the limited-selection approach can't work for everybody. Another is that, as my Bloomberg View colleague Megan McArdle wrote last year, if every retailer paid like Costco, many of Costco's labor advantages would disappear. And finally, while some retailers surely have hurt themselves in their zeal to optimize labor, the move away from full-time retail jobs and toward staffing that's closely matched to customer demand hasn't been totally irrational.

In one recent study, Saravanan Kesavan, Bradley R. Staats and Wendell Gilland of the University of North Carolina looked at labor practices at a large (unidentified) retail chain. Their hypothesis was that the use of temporary and part-time workers would be linked with per-store sales in an inverted U-shaped curve -- with sales at first rising as the percentage of temps and part-timers rose, but eventually falling.

The data backed them up. To maximize sales, the optimal share of temp workers was 13 percent and part-timers 44 percent. But those percentages were both higher than the retailer's current averages of 7 percent and 32 percent. Overall, hiring more part-timers and more temps was likely to lead to higher sales.

The data-driven reexamination of labor practices by big retailers will surely lead to some improvements in how workers are treated and paid. I don't get the impression that, by itself, it will lead to all retail jobs becoming good jobs.

Source: Bloomberg