I’m sure half of you (or more) have listened to the “Comcast call from Hell”. For those that haven’t (and can’t be bothered to click through and listen to an 18 minute painful phone call), basically here is what happened:

A guy tried to cancel his Comcast service. The agent on the line wouldn’t let him do it. No matter what the customer said, the agent would turn it around and stop the cancellation from going through.

It effectively turned into a conversation like this on repeat:

“Please cancel my service.”

“Why would you want to cancel your service? Comcast is awesome!”

Painful to listen to, and likely even more painful to experience.

And sorry. It’s my fault. At least indirectly.

I spent a good portion of my time at McKinsey specializing in how to set-up, manage, and optimize “Save Desks”.

What is a “Save Desk”?

A Save desk is a specialized call center that only takes calls from people who are planning to cancel their service.

Fifteen years ago Save Desks didn’t exist. Instead call center agents were generalists. They would handle new sign-ups, customer service calls, and process cancelations (and anything else that caused customers to call-in). The problem was that call center agents are very low paid but very expensive at the same time (since companies need so many of them). Trying to train a low-wage employee with high turn-over to do ALL the jobs was difficult. And it was even harder to measure them on the success of all those jobs. For customer service calls you wanted the customer to leave happy, and do it as efficiently as possible (i.e., customer satisfaction; handle time per call). For sales calls you wanted to incentivize sales (and you are likely ok if handle time was a little longer if it resulted in a sale). For cancelation calls you wanted to just keep the products the customer called in on.

When all the functions were done in one group, saving customers was lost completely. It was easy to measure sales. It was easy to measure handle time. But a lot harder to measure when you saved something – there was no way to know someone was calling to cancel unless the rep told you they were. If you let the rep make that decision themselves and then evaluated them on ‘save rate’, then it made sense for them to claim every call as a potential cancel. All of a sudden they would have 99% save rate.

And the skills were different. Being good at solving customer complaints was different from knowing the features and upsell offers you could make. And both were different than dealing with a customer who wanted to cancel.

The result was companies were terrible at saving canceling customers.

My job was to fix that.

Saving Customers

I traveled the world helping telecoms (just telecoms – my peers focused on cable companies so we weren’t sharing secrets with competitors) create save desks from scratch. And we had a system.

We would create a separate group that only took cancelation calls. The rest of the organization wasn’t allowed to cancel anything – they had to transfer customers who wanted to cancel to this special desk. That desk was given special training and coaching. They did daily huddles where they talked about different techniques. They were managed by specialist managers. For metrics we ignored handle time. Their core metric was their “save rate”. To calculate save rate we used some funky math:

{[Total calls handled] – [Total products cancelled]} / [Total calls handled]

So if you took 100 calls and half the calls canceled 2 products, you would have a 0% save rate. And yes, it could go negative.

In addition, for every call the agent was required to record what the reason for the cancel was. With that data we could measure the save rate based on reason. Turns out when someone is moving out of the country the save rate is basically zero for example (actually very negative since it’s all their products they are cancelling). But these numbers let us figure things out. We learned what the major reasons were people were canceling and what reasons we could save people and which reasons we couldn’t. We also learned which agents were good at saving which reasons, so we could learn from the best and reapply to everyone else. We could do targeted coaching: “Sarah, you are doing great, but when women from New York State call and try to cancel because they are moving to cell-phone only, your save rate is really low. Let’s go through some techniques that work really good in that situation.”

My favorite specific example was for the reason: “My son has his own line and now he’s moving to college, so we won’t need it anymore.” The save rate on that reason was close to zero.

So we solved it.

We created a special product that was only available at the save desk. It was only $5/month. It allowed you to have a separate line that only took inbound calls most of the time, but it could make outbound calls on weekends and holidays (including a 2 week time period around Christmas). It was designed so that when little Jonny came back from collage on over Thanksgiving he wouldn’t tie up the main house line. Save rate went from zero to almost 100%.

At one point when I was considering my next step after consulting I considered becoming a save desk specialist. My idea was to serve non-telecom/cable companies who had subscription products and help them set up their own save desks. Magazines and newspapers couldn’t afford McKinsey, but they could afford me if I went direct. I ended up going a different route with my career, but I still think that niche is viable for the right person.

Creating a save desk creates a ton of value for a company.

It can also piss a lot of customer off who just want to cancel their service. The problem is, from a company’s point of view, by the time you are ready to cancel, pissing some people off a little bit more is totally worth it if you can save 50% of the others. That’s a lot of revenue you hold on to.

So: While that Comcast agent was a little over the top (a lot over the top), he was just doing what I had incentivized him to do. Maybe next time I would put in a score around customer satisfaction. Or maybe just handle time. But if I did, I would want to test it. I’ll bet it would destroy value vs. just keeping it the way it is and dealing with fallout like this whirlwind every half-decade or so.