When is a call a complaint? Can language change the whole way you look at something? Glenn Elliott thinks so.
Over ten years at Reward Gateway we used lots of metrics to track internally how things were growing. We were always upbeat and sharing internally the good news that we were on a steady and continuous growth curve.
One of these metrics was “Number of Enquiries” which was the number of calls, emails and instant chat sessions our user help desk had handled. As the company grew, the number of enquiries grew and our customer service team grew accordingly.
Every quarter at our all-hands meeting we celebrated the success of this hard-working team in their amazing feat of handling all of those enquiries. They were our heroes – handling the hard work that inevitably (we thought) came with the success of our sales teams and customer retention teams. The helpdesk team met their SLA’s, got great reviews from customers and… cost us more and more each month.
Then, someone pointed out that another way of describing these “enquiries” was “complaints”.
“No one actually wants to call the helpdesk and ask for help… The best service is actually no service. If people could do what they wanted to without needing to contact us at all, that’s what the best service actually looks like.”
That was a new way of thinking about something we had actually been thinking of as a growth metric.
We’d been looking at it all wrong
We thought great service was picking up the phone in four rings, solving the problem and getting a good rating after the call finished. No. This person was pointing out that the best service was no need to phone us at all.
When we realised our mistake, we had to act. We re-named “enquiries” to “complaints” in every internal report, message and piece of communications. Even the most gentle “How can I do X?” enquiry from a customer became a complaint in our stats. This action meant that we could see these interactions for what they really were – something that the user didn’t really want to do. Our product and service would be better for our users if we worked to reduce the number of these interactions rather than accept them as part of our growth.
Once we saw enquiries for what they were, we could act to minimise them
Another helpdesk employee pointed out that whilst we tracked data on all of the call types and we knew exactly which parts of our product caused phone calls and helpdesk tickets, that information never got back to the product managers and engineers responsible for those products. We had an incomplete feedback loop! All of the data but not in the right hands.
Quickly, we implemented a metric called DPMO – ‘Defects Per Million Orders’
DPMO is simply a ratio of how many defects (or phone calls and complaints) in our world are created for each million attempted transactions.
DPMO tells you how much trouble each product or system component is that you have. The trick in implementing DPMO is to be ruthless in calling everything that is not an intended user outcome a defect.
So think about system login – your users want to log in, don’t they? That means every failed login, forgotten password, every blocked account, you count that as a defect because it isn’t what the user wanted. Then you get that information back to the team responsible for that part of your system.
So the team responsible for user login gets to see how many times per million passwords and logins get rejected – even though for most of these, the user never raises a real complaint.
Task the team to reduce that number every week or fortnight, and you start to see amazing things happen. Rather than giving engineers tasks and projects, you give them ownership of a key metric that their code affects. They start to make changes you hadn’t imagined and get excited about measuring the success (or not) of their latest code change.
It completely changed how we looked at a key metric – a metric that had been staring us in the face for years.
The end results were better service to users who needed to call us less and lower costs for the business. Truly a win-win.