Survivorship Bias: Don’t ignore the losers

Vincent Frisina Profile Image

VINCENT FRISINA

Director, Customer Marketing & Programs

I bet you don’t know just how helpful failures can be. It’s pretty typical for humans to learn about the world through failures. Touching a hot stove, falling out of an unsafe tree, or walking through a burr-patch with bare feet all present clear opportunities to learn from failure in a painful way. The human brain is programmed to learn from the pain, and avoid the situations in the future. But what about times when the pain is less pronounced? Are there groups of failures we should be revisiting, but aren’t?

We’re exploring the Psychology of Telecom Expense Management with a look at Survivorship Bias. Our brains are pre-programmed with these biases whether we like them or not. Through understanding, however, we can avoid certain categories of mistakes and potentially deliver better products and services to our users, while also benefiting our bottom line.

Survivorship Bias depends on our tendency to use evidence of success as the primary measure for planning for future successes. You may recognize it as a kind of sample bias, or how some data points get ignored because they don’t prove the theory. Laszlo Bock, head of People Operations at Google, discusses how survivorship bias impacted Google’s hiring practices on an episode of You Are Not So Smart. After years of tricky hiring practices and insane interview questions, the human resources teams at Google decided to use some statistical analysis to see just what a great job they’d done.

“When failure becomes invisible, the difference between failure and success may also become invisible.”

Here’s the problem. Although the interview process predicted the candidate’s failure, Google never hired the person to find out if they actually failed. All measures of success were based on performance of employees who made it through the interview process successfully. Google went back and hired some of the washed out candidates. The interview washouts performed as well as the successful interviewees. Google discovered that while its hiring process was newsworthy, it did not actually predict success and that by eliminating failure data from its sample set, it made the difference between failure and success harder to see.

One place “survivorship bias” rears its ugly head in our world is in zero usage device management. Unused and active device counts certainly drive unnecessary business expenses each month. Telecom Expense Management (TEM) and Wireless Expense Management (WEM) firms focus on keeping this key performance indicator (KPI) down. We consider the zero total usage metric our success pool; if the number is very low, we have a high success rate in keeping zero total usage devices down. What about the lines with only 10 minutes of usage in a billing period? How about those with just a few kilobytes of data usage? These lines have survived the culling of unused lines, but the logic used to select the sample is biased.

With Visage Mobile’s MobilityCentral product (shameless plug), you don’t have to choose between analyzing your successes or the failures — MobilityCentral will do both for you! Zero usage metrics on the Spend dashboard deliver instant access to candidates for disconnection, and canned reports easily surface lines with minimal usage in various categories. MobilityCentral also keeps track of contract and termination fee information for every line, so you can see if it’s cost-effective to cancel any active line of service.

It’s always important to review the criteria used to determine your successes. When you know what your failures can tell you, you can better see success!

Managing wireless data pools is like playing the Price is Right

Telecom expense management encompasses a huge array of communications tools. One of those tools is wireless data connectivity. Mobile data continues to be a critical aspect in how employees get their jobs done. Determining how much data to buy each month can be a time-consuming and laborious process based on a previous month of billing and usage data. How do you hit a moving target each month to ensure you’re buying just the data you need without going over?

International Savings

International roaming expense come in a variety of flavors. Similarly, international savings opportunities abound when you know what to look for.

The Politics of Mobility Management

People-pleasing. Job-protecting. Flying under the radar. These probably aren’t phrases you’d expect to use when describing mobility managers, yet these attributes are more prevalent than you’d think.

Why Telecom Expense Management Matters

VINCENT FRISINA← Back to blog home The types of costs and expenses we incur directly relate to the products and services our businesses provide. Telecom expense management encompasses the tools and methods used to minimize costs and allocate expenses to the assets...

Why Mobility Managers Should Care About “Orange is the New Black”

What’s the number one reason why you should care about the new season of “Orange is the New Black?” Because your USERS care.

Why “Pokemon Go” is catching mobility managers unaware!

Pokèmon Go launched a couple of weeks ago to much fanfare. But what does this game have to do with how you manage your corporate wireless program?

Distracted Driving

We all know distracted driving is on the rise, shouldn’t be done and has the potential to be lethal. Yet how many times have you, or someone you’ve seen, read a text message, placed a call or used an app while behind the wheel?

T-Mobile DIGITS Reinvents the Way You Buy Wireless Service….Again

T-Mobile recently opened their DIGITS service to a public beta test. The service allows mobile numbers to be used on multiple devices at the same time, or for a single device to have multiple mobile numbers.

MobilityCentral Release 5.7

MobilityCentral Release 5.7 is here. Take a look at the newest features!

More and More Data Usage: How Much is Enough?

Larry Treas discusses one of the primary areas of charge-blindness in a mobility program: data usage.