Today’s contact center is much different than the call centers I worked in some 20 years ago. At that time, the only channel was phones and the systems were bulky and required highly skilled technical support people and engineers.
Now we have out of the box contact centers that support omni channel engagements and can be setup by anyone who can use a computer. Setting up these systems isn’t challenging, but maximizing their performance definitely requires a solid combination of gut instinct and experience.
Once your system is built, how do you know how your employees are performing? This question perplexes “out of the box” all the time. While the system can give you readings on quantitative performance, they leave much to be desired. The omni channel systems may include email, live chat, phones and sometimes even social media, but there are very few that can actually deliver the work to agents based on their skill and availability. In fact, I know of only one.
If this is the case, how do you measure KPIs and performance? We will look at this two ways and if you don’t have a data scientist on staff, time to get comfy with trusty Excel or Google Sheets.
Deciding on Key Performance Indicators
You’re going to measure KPIs in two parts: Customer Experience and Agent Level. In this post, we will focus on Customer Experience Metrics only.
Customer Experience KPIs are the KPIs that, well…. Speak to, wait for it…Here’s the reality, if your customer service isn’t fast, customers will become furious. Always bank your Customer Experience on efficiently resolving Customer issues (do it right, do it fast). your Customer’s Experience.
KPIs that will help you determine if your service is efficient are:
Ticket count (Inventory) and Backlog
Ticket count references the number of tickets in your queue to be worked. There are varying opinions on what to include, but I tend to include any tickets that are awaiting a response from an agent. If you’re using Zendesk, this can include open and on hold, as I always advise clients to use pending status when you’re waiting to hear back from the Customer.
Though it seems simple, keeping track of tickets to be worked can be quite tricky and in some cases tickets can be unaccounted for. This is why tracking your backlog is so important. As George Xourafas, a Workforce Management Expert out of Ireland says, your back log is the “age of oldest item and average overall age of inventory.” Inventory here refers to the number of tickets waiting to be worked.
Response time encompasses both speed to answer (sometimes called wait time) and resolution time (time the inquiry was responded to and resolved by an agent). Why not just speed to answer? Some systems will automatically respond to acknowledge receipt of the message. That’s greatly skews the time. You’re looking for a completely resolved engagement.
Number of Responses before resolved
This one is a bit tricky. I’ve heard lots of people focus solely on FCR (first contact resolution), but this sometimes leaves you with a novel for a response instead of the facts. This also kills your handle time (more on that next). Use this metric as directionally sound information. For example, about how many responses does it take to resolve a customer’s issue when the request is regarding “x”? You can look into the responses and decide if perhaps a modification is needed to your current response, or a self service process should be put into place.
This metric is used to measure how long it takes to type and send the response. Some companies force their employees to meet a target. I’m not an advocate of that experience. Instead, what I think works best is to watch the average of the team and pay close attention to outliers. Look at your most ideal workers, see what they’re doing and implement that as part of a training process. For me, handle time is more about planning and tells me if my employees actually understand the work they’re doing. You can learn a lot just be viewing how long it takes to respond to a Customer.
Speed to answer or Service Level
Also known as Average Speed to Answer (ASA), or time it takes to respond are solid measurements. When you call someone on their cell phone, the wait time before voicemail picks up is 30 seconds. For me, 30 seconds is a fair average speed to answer time (of course this varies by industry). You may also opt to measure a service level. Service levels speak to a percentage of calls answered within a set timeframe. For example, 80% of calls answered within 20 seconds is a widely used metric.
Call abandonment Rate
Call abandonment rate measures the percentage of calls that go unanswered after being delivered to your queue. Delivery to queue means that the Customer has listened to your IVR (interactive voice recognition) options and is waiting for a live agent to take their call. If the Customer hangs up during this process, it counts towards an abandoned call. There’s a strong argument to exclude any calls not delivered to queue (abandoned in IVR), as the reason these calls have dropped are less likely related to long wait times. Usually the higher the speed to answer, the higher the abandonment rate.
Not Ready Time
This one will be called very many things depending on the system, but it’s important to understand the measurement is simply this: the agent is not available to work this channel. The reasons can very: meetings, trainings, after call follow up (notating account, outbound calls, etc.). Reducing your Not Ready time will increase your productivity time. But keep in mind, Not Ready time is necessary and should be planned for via shrinkage.
It’s important to measure repeat callers within 7 days (may vary by industry). This should be a rare occasion, but ultimately, you’ll want to know if the person is calling for the same reason. Checking into these will help you identify training issues, broken processes and other opportunities to improve.
Handle time is also quite relevant for phones (see above).
Speed to Answer
Live chat works best when the responses are in as close to real time as possible. A rule of thumb would be to avoid having the Customer waiting for you to respond for more than a full minute. In this case, you can engage the client with automatic responses that will provide quick updates for them. I don’t know about you, but if a chat gets to quiet, I’ll move on to something else and forget I was even in conversation.
Chat abandonment Rate or Incomplete Chats
With this metric, you will learn how many chats were not completed. I’d also take it a step further to see if the Customer used a different channel to resolve their issue. Similar to repeat callers, you’ll learn a lot about how Customers are going about getting their issues resolved.
Total Chat Time
If agents are handling multiple chats at once, you may find that the chat time is longer on average. Decide if this is the experience you want to create, or would it be more beneficial for the agent to handle fewer chats and resolve issues faster? I vote for the latter.
Other measurements you’ll need across all channels are Quality Assurance (best if done by a neutral party), and of course a Customer Satisfaction score. Most out of the box contact centers will include a way to gather CSAT feedback from your Customer. What’s most important is not to take a cookie cutter approach. Listen to your Customers, they’ll tell you what they will tolerate and what’s not acceptable.
I’d love to know your thoughts.