Automation has become synonymous with “cost savings.” Cut heads, streamline processes, get information sooner/better/faster; it will save you millions. We’ve all heard it, and it’s not surprising. There have been phenomenal advances in technology over the last 20 years which allow companies to collect and aggregate data into meaningful analyses for operational decisions. Data entry and processing jobs have been replaced by software and retail cashiers have been replaced by self-service check-outs much like artisan weavers were replaced by power looms. While some fear the short-term unemployment of machines replacing people, technological unemployment is one of the best things to happen in a society. Anytime a group of workers is replaced by technology, they are available to contribute something else to the economy, to solve new problems, to cure diseases.
The point of this essay is not to deter anyone from investing in automation and technology, but to remind executives that technological investments need to be evaluated the same way you would assess any other monetary decision: does the investment cover your hurdle rate? What risks does automation bring, and how costly are they?
There is a lot of judgement in calculating the savings of technology initiatives since companies rarely have exact measurements of total cost. The time your employees spend on tasks is a necessary input to calculate internal rates of return, which is difficult to measure accurately. Take a simple example: suppose there are 3 people in your accounts payable department who cost $10,000 per month and spend a third of their time reconciling monthly invoices to bank statements. You learn of a new software that can perform the exact same function which costs $1,000 a month. Great! You can reduce your staff by one, lowering payroll burden to $6,667, add this software for $1,000, and save $2,333 a month. After the software is integrated, you learn it still requires some manual inputs and full detail reviews, which prompts complaints from your AP department that it now takes up 40% of their time.
You now spend $11,000 a month on AP and have unhappy, overworked employees. What went wrong? The first common mistake in automation decisions is identifying the specific task you are automating. The AP department said they spend a third of their time on bank reconciliations; ok, but doing what? Is it literally tying out the invoices to the bank statements? Or is most of their time spent researching discrepancies and following up with outside vendors? The latter is probably right, and in that case, automated software cannot replace anyone.
The second common mistake is not appreciating the limitations of technology. Software does not know anything you did not explicitly tell it. New vendors need to be input, payments need to be accurate to the penny, and your new AP software does not know that ABC Incorporated LLC is the same vendor as ABC Inc. LLC. The software gets you 99% of the way there, but that 1% takes 99% of the time. Close only counts in horseshoes and hand grenades.
Aside from the cash out the door, costly inefficiencies transpire when the human thought process is removed from the workflow. We are living in the age of an unhealthy reliance on technology: people trust their weather app more than sticking their hand out the window (“It can’t be raining! My app doesn’t say that”); Uber drivers trust their navigation only, never their customers (“No ma’am, you ARE home”). People have come to trust operations software in the same way. Monthly reporting and daily operations management have been outsourced to machines with unquestioning trust. Machines don’t make mistakes, people make mistakes. Well, everything machines know are from people telling them. “Artificial Intelligence” is still just a bunch of rules, created and refined by humans.
Technology has glitches, crashes, the equivalent of a human ‘fat finger.’ If you set up a system correctly and have the right checks in place, it will usually give you accurate outputs, but employees still need to take the time to ask, ‘does this make sense?’ When employees blindly rely on system reporting, it doesn’t just hinder operational analytics, but also detaches your workforce from accountability. Your employee didn’t make a mistake, the report did. This lack of ownership and responsibility results in costly delays and financial black holes.
Despite all the points raised in this article, technological automation is usually a cost savings. However, let’s not go overboard. Just because it says ‘automated’ and ‘one touch’ does not mean it is more efficient for your organization and better for your bottom line.