May 06 2014
Lessons in Business Intelligence: Be Careful What You Wish For
Author Note: This post was originally a guest post published on “The Decision Factor” blog. Since that site seems to have disappeared I am re-posting it here. The original publication date was November 1, 2012, but I believe the content is still relevant.
What’s the purpose of a business intelligence (BI) dashboard? It’s not just to look sexy. The primary purpose of a dashboard is to convey information. A secondary purpose is to inspire a behavioral change based on the information that’s being conveyed. Nobody wants to be “red” on a dashboard reviewed by executives, so they’ll change their behavior in order to get into the “green” area.
But humans are a creative species. What happens if their behavior changes in an unanticipated way?
Unintended Consequences
Perhaps you’ve heard the story of the cobra effect, dating back to the time of British rule over colonial India. The government was concerned about the number of cobras in the cities, so they set up a bounty program that paid for dead cobras. This provided a financial incentive for the general population to hunt snakes, with the intended result being a decline in the overall population of snakes. I doubt they had a dashboard, but I can imagine what one might have looked like…
I think what actually happened is a wonderful example of human behavior. Rather than spend all their time and energy trying to hunt down the wild cobras, enterprising folks started cobra farms! It seems that the financial incentives made that more profitable than catching and killing wild cobras. Since this wasn’t the intended outcome, the government cancelled the incentive program rather than try to determine the origin of every dead cobra turned in for the bounty.
What happened next is the real punch line to the story: having no more market for their product, the cobra farmers turned all of their snakes loose. The net result: far more cobras in the cities than before the incentives were put in place!
If we’re not careful, the same thing can happen to our best efforts today. We’re still human, after all.
When the Dashboard Backfires
I heard this next story from someone at a manufacturing company. They had determined they were losing sales because they couldn’t manufacture product according to plan, mainly because the warehouses would occasionally run out of raw materials. The next step seemed obvious—set up a dashboard to monitor out-of-stock situations. A dashboard was created, and sometime later the company congratulated itself on solving the problem. The dashboard was completely green; no out-of-stock situations had occurred for many months.
And then the cobras were set loose.
Warehouses were given incentives to avoid being listed in the “red” area of the dashboard. The intended result was for the warehouses to do a better job at managing raw materials. Instead, the warehouses simply started stocking more raw materials than were required.
The incentives made it imperative to avoid out of stock situations, so they did. Since the warehouses ordered more than necessary, they were never out of stock. Did I mention yet that the inventory being produced was perishable? And so were many of the raw materials?
The dashboard showed the truth: there were no more out-of-stock situations. However, the company still had to discard a substantial amount of raw material as it expired in the warehouse. The dollars lost due to the expiration of raw materials far exceeded the cost of the original lost sales. The dashboard solved one problem but created a far worse one.
The moral of both stories? Be careful what you wish for—you just might get it. In these cases, the road taken to get results was more expensive than intended.
When it comes to dashboards, consider the behavior that your dashboard may inspire, and try to anticipate all the possible outcomes—just to be sure you’re not releasing a bunch of snakes.
To learn more about the economics of the cobra effect on financial incentive schemes, check out The Cobra Effect podcast on Freakonomics.com.
Great thoughts! Perhaps the point here is that suddenly measuring & reporting on incentives that are already in place might actually demonstrated whether those incentives are, in fact, in line with where the organization needs to head. It seems to me that the next step for the inventory case is to further refine the inventory incentives to balance out of stock conditions with carrying costs.
Anyhow, great examples!
This is why you can’t look at your data in a silo. When you change A the effects will ripple out all the way to L. Nothing happens in a vacuum! Dashboards are great because they make it easier to manage big chunks of data, but when humans are sill in charge of interpreting that data mistakes can and will happen.