We can use the current SARS-CoV-2 virus outbreak's impacts on businesses and labor forces as a test case to assess the impacts of a remote labor force on overall production. As is the case with any failure, one is presented with the opportunity to learn. While some businesses are less impacted than others, there is a dramatic change occurring globally. With the increase in preventative regulations, businesses in many forms have been forced to close doors or make dramatic changes to structure. At home labor has had Sysadmin channels lighting up with calls for hardware and additional IT support staffing. Some companies are well enough off and have been rolling out this kind of work archetype intermittently but for others this is a foundation shock.
At home labor is impacting the way you manage your team, how they manage themselves, and how the company manages incentives and performance monitoring. This likely has caused some major issues, as any abrupt change does. Can it be a learning experience for the more dynamic workforce of tomorrow? How can you establish a test environment your labor force? One that can see little to no impact in GDP or company output when something dramatically changes the way we live our lives. I think it daft to assume these events would occur in any less regularity as time goes on.
Lets create a test case with labor. One that allows us to manipulate variables individually while also implementing a reward system. In this case it is easy because the subject is in most cases viewing remote work to be a reward in itself. We can use the productivity functions below to select our most productive units and reward based on productivity standards. E.g. Project assigned/completed time metrics.
I. A Test Environment
you may have heard of A/B testing or a multitude of other techniques but essentially what needs to be done is:
1. Establish a consistent measurement mechanism
2. Control for external influences
3. Use the structural force to at home labor as our change case
4. Compare data and test hypothesis
1. Measurement
You may already have a metric for your labor productivity but for those who don't it is a function of labor cost and labor output. This may be a revenue value for sales/marketing teams or a quantity handled value for support and ops teams. Accounting may look at batches processed and the list goes on. The point being that the information is out there and most of the cases i just provided come with software tracking some version of metrics i mentioned. You could also design your own, in our test case:
Managers rate productivity of workers 0-100 as ProdM in the graphic below.
Once you have a KPI and a supporting variable information such as per laborer: Pay, time with company, hours worked, average output, total output and any other variables you may consider to impact or explain labor output. We can catalog them in a data format which we know will be consistent for our use. Automate and use databases if you have them but this could easily be done in excel or .csv as well. You'll want to make sure each record is time stamped and a daily record level is fine unless you're dealing with larger quantities.
2. CONTROLS
I always suggest adding some level of classification to the data set such that each record holds a reference to a division, segment, or unit level grouping. This will greatly help in understanding and controlling for differences in group's or region's later on in visualization. We can use the segments to understand outlier cases and effectively grade the labor force such that you could list employees by productivity(x)= Output - Input for a given time period. This should be done for one case prior to working from home and monitored ongoing throughout this situation, if nothing else you have an impact analysis case.
With proper segmentation we can see the productivity per salary dollar as it flows through the different departments.
3. Change Case
In most experiment cases you control and then change the variable you want to test. In this case the variable we have is: at home vs. in office; or whatever your starting case is prior to COVID-19 impacts. If you don't have measurement for your starting case then you will have to have a sample of your total work force come into the office or return to standard work activities while leaving another part of your set at home to give a A/B state to compare as regulations subside. Categorize these in their own subgroup for differencing.
4. Compare and test
Once you have cataloged the required info you can analyze the differences in both at home and in office states to extrapolate whether your labor is more productive remote. Remember that its more than revenue, its efficiency, its employee happiness, and its likely the future of a global workforce as many industries are moving in that direction.
II. Conclusions
Play with the data, but understand tests for significance. Averages and outliers can hold lessons for managers but be mindful an employee could have been dealt a better or worse hand or less productive variable segment. In our case we learned that ops should be using more revenue tied incentives and less base. In this case Ops deals with unit after use turnover rate, an increase in efficiency there would boost the quantity we could supply. The employees who were already remote will self manage better by directly relating their time to compensation in the form of a commission or bonus tier of income. Let managers adjust the % a tad and you get a self firing feedback loop. If you work in IT hardware and have ever wondered why laptop over desktop, when desktop is the clear front runner in capability/price. This event is an obvious example but what about phones, or dare we say cloud computing for the deepest of pockets. Individual desktops can be converted into terminal servers, allowing people to do intensive video editing from a chrome book. The fact is we work better with compute mobility. So whats the play? Maybe nothing; some industries like Hospitality, or other service types need to be in person to do business to some degree. For the business segments with employees that have been forced home and can still work. Why not capitalize on everyone being home and begin to understand your employees better instead of potentially disrupting your flow of business by testing sending people home in a time where there is no need.
I'm interested in applying AI and Data Science to world problems. If you have a research topic suggestion or would like to talk about projects please reach out! Here