Meanwhile, Walmart can increase profits by simply hiring fewer workers (or laying off more). Allegedly, in 2012, Walmart was spending $12MM per second of transaction time in cashier wages alone[1].
That's per second of average transaction time, meaning one second times the total number of transactions (per year I guess?). Kind of a meaningless metric in my opinion.
Good point, I glossed over that. I'm not sure I understand your interpretation properly, though. My interpretation was they took all of the transaction times, divided them by the total number of transactions, and then divided that number by their total expenditure in cashier personnel costs. I wouldn't go so far as to characterize this as "meaningless," but I will admit it's a little weird to measure your personnel costs in terms of dollars per second of the average time a given transaction takes. I think the idea is to shock people with a big number -- 12MM per second is shocking. However, the article doesn't explain that Walmart employs more people than any other company, world-wide[1] -- that's also a big number, and so when you divide the two, a much smaller number appears. The more interesting number, I think, is the total personnel costs versus the number of personnel -- some average or quantile that demonstrates, on a smaller scale, the ratio of personnel cost. Now, multiply that number to scale back up to number of transactions and you'll get a better representation of the personnel cost per transaction, without the oddball time axis.
In fact, as I'm thinking about this, perhaps you're saying it's a meaningless metric because it makes no sense to compare dollars per average transaction-second to dollars for all cashiers, as the units aren't correlated and just give random values.
[1] http://www.reuters.com/article/walmart-idUSL2E8E72FF20120307