The Failure of Self-Service Automation

Enabling self-service for employees and customers is today’s big efficiency play. It can reduce costs and, when done well, streamline and speed up processes. Two-thirds of IT leaders expected to increase self-service enablement in 2019, and within two years, broad-based self-service deployment will increase by 2.5X. The Hackett Group benchmark study uses the degree of self-service automation as a criteria for world-class performance – world-class IT organizations automate twice the number of eligible processes compared to peer group IT functions. But there are huge risks to self-service that have some CIOs backing away or at least slowing down the full-court press.

First, although most business leaders won’t come out and say it, their self-service automation shifts what was once their burden onto the shoulders of employees and customers. If the steps of the self-service process are clunky, repetitive or prone to breakdown, the burden will alienate workers and customers and degrade trust in IT, the process owner and/or the company. Customer-centric design means just that – the system should be architected to work the way a customer would expect and appreciate. Too often, that isn’t the case.

Second, while a great deal of attention may be directed at user interface design or data integration with transactional systems, do process owners and their IT collaborators consider how the user will get help when (not if) the system fails? Can a frustrated customer easily access a human for help, and can that human see what the customer has already done on the system (data entered, etc.) in order to efficiently diagnose and resolve the problem? For too many companies, the answer to these questions is no.

I had two experiences this summer with very big companies that should know better. In both cases, the self-service technology failed, and both times I was unable to get adequate help from the company. (Rant begins here.)

I didn’t receive a package at my house, despite the carrier’s tracking history showing it was delivered. The carrier had an online self-service system for filing a missing-package report. I have to log into my account and fill in a dozen fields of data about myself, the package, and the company that sent me the package. Just as I finish, the system crashes. I have to reenter everything from scratch because the system can’t retain my entries, despite the fact that I’m logged in. After multiple crashes and reentries, I hunt fruitlessly on the site for a customer service contact. There literally was none. However, a Google search turns up a number.

I tell the carrier’s phone representative that the online form isn’t working for me. He asks for all the data I’d been entering, puts me on hold, and concludes the call by telling me to look for an email from the company (you already know where this is going). The email arrives, I click on it, and it takes me to the blank on-line missing-package form that began this whole excursion in self-service futility.

On different hot summer day, I needed to rent a small truck for an hour to transport some doors I would be purchasing. The truck-rental company now offers the option of bypassing the rental counter and unlocking the truck with my phone. Cool. I create an account and opt for mobile pickup. A video explains the series of steps this entails and promises it will only take a few minutes.  I drive to the truck lot, find my truck (locked), inspect for damage, and use my smart phone to upload a photo of my insurance card, license and a selfie so they know I am the same person who booked the reservation. The system churns and churns as I wait. I’ve been in the parking lot for 20 minutes.

Bad news: the system fails to do an accurate face recognition match. I am prompted to provide additional forms of ID verification. This includes, incredibly, the phone numbers and emails of two people who can verify my existence. Clearly, the company’s legal and risk team had a lot of influence on this system’s design. I input my wife’s and brother’s information, not knowing if they would respond to an unexpected inquiry from a company asking about my personal identity. Would they even see the request? I frantically text them both, imploring them to respond. The self-service screen continues to show “waiting for response.” Thinking the system might be stalled because of my weak phone connection, I wander the parking lot seeking a better signal. The churn continues.

Back in the car,  I start a chat, tell the representative I’ve been there for more than 30 minutes, and ask if they would like an employer letter of reference or a bank statement for additional verification. The representative responds in three-word sentence fragments. He/she apparently doesn’t have access to see what is happening (rather, not happening) on my phone. After a few mutually confused exchanges, I learn that I have been validated, but since the system isn’t showing that on my phone, I can’t unlock the truck, and neither can the representative. I abandon the chat and go home, where I cancel the reservation and leave a comment to the effect that this is the worst-designed vehicle rental system I have ever experienced.

The lessons for IT and customer service teams are many. Data entered by a logged-in user should be retained (temporarily) and presented even after a system crash. Security features should be reasonable and not depend on unsuspecting third-party’s availability and cooperation. Mobile systems should function even with one-bar connectivity. But most importantly, the human-based back-ups to automated self-service should be easily accessible, properly trained and integrated with the system so that they can see and respond to the problem. Until we get that right, technology-based self-service will continue to put companies at risk.