Neil Irwin had an interesting Upshot piece highlighting a new paper by J.W. Mason arguing that slow productivity growth is in large part due to slow GDP growth. The basic argument is that if growth were faster, labor markets would be tighter, and companies would have more reason to invest in labor saving equipment.
While this argument strikes me as undoubtedly true, there is another aspect to productivity growth that is often missed. One thing that is even easier than replacing workers with equipment is simply not replacing workers. In other words, most employers can run stores, restaurants, or other businesses with fewer workers. The cost of this is likely to mean that customers have to wait longer to be served.
This could mean, for example, that when you get to the checkout counter at a supermarket you have to wait ten or fifteen minutes in line rather than having someone immediately available immediately to serve you. The same would apply to lines at fast food restaurants or the pace of service at a sit-down restaurant. Instead of having workers available for customers at all times (which means they do nothing, some of the time), employers will make customers wait.
This would show up as an increase in productivity as conventionally measured. Output would be unchanged, but fewer workers are employed than in the good service scenario. In principle, if we have perfect productivity data, this would not be the case, since the longer wait times should be reported as a deterioration in quality and therefore a price increase, which would mean lower output. But we don't have perfect data, so in our productivity numbers, longer wait times mean higher productivity (and vice versa).