All too often the cost savings from efficiency improvements of various kinds are dramatically over estimated, especially when large numbers of small improvements are aggregated.
Companies of all kinds (not just Pharma) often justify investments in new equipment or software based on predicted time savings that result from the purchase. “If we buy a new AutoStamper-2000 for $10,000, we’ll save 1000 hours a year in applying postage stamps. That’s $50,000 in salary costs, about what we pay one employee, so we can reduce our headcount by one.” Sounds like a no-brainer.
But… what if those predicted savings are based on an estimate of 1000 employees saving 1.2 minutes per week of stamp-licking? Is that cost savings really recoverable? Will those weekly minutes actually be re-deployed on a more useful job-related activity? What about the potential additional cost of those 1000 employees walking 3 minutes every week to use the AutoStamper-2000 on the 3rd floor? And was the full cost for a new AutoStamper maintenance guy included? Sounds like a silly example, but this sort of thing happens all the time, whether it’s implementing a new piece of automation, a new contract employee (who will do “low value” work so your high-paid employees don’t have to), or a new outsourcing relationship. In the real world, I’ve seen proposed numbers like 15 or 30 min a week of time savings, summed over a few hundred employees. It’s almost certain that less than 100% of that time is actually meaningfully redeployed on work. Saving an hour or several hours a week per person — now that’s real time savings. But not 15 min per week.
Here’s another related example. Companies often tally up the fully loaded cost per employee, including the cost of their benefits, office space, computer, and other allocated overhead. In industrial research, the laboratory reagents and supplies that an actively working scientist might use are also sometimes included in the total. Things can go wrong however, when you use that entire cost number to calculate the savings from using more automation or cheaper contract employees. The problem is that a lot of those costs might be fixed, and won’t actually go away unless you close a building or make other major changes. The new contract employee still needs a computer and an office. The research building doesn’t get any smaller, and the same reagents will need to be purchased if the same research projects are active. As a result, the true cost of the entire operation is not reduced as much as expected — and in the worst case, costs might actually rise.
Bottom line: Make sure calculated cost savings are actually recoverable, especially if you need to spend real money to make the proposed improvement!
MechanizedIntelligence 