The 6 KPIs for Improving Your Live Chat Efficiency
A focus on KPIs can really boost your chat performance. But looking at our own Analytics, I know that all the tracking options can feel overwhelming.
Being number-driven is great, but there’s also the risk of paralysis by analysis . Before you can draw measures from the numbers, you need to know what to measure.
Here at Userlike, we are heavy users of our own live chat software to support our customers. Let’s look at the KPIs that we find most important for our reports, fill the numbers with meaning, and watch for common pitfalls.
The number of chats you get is the first metric you should look at to get an overview of your chat’s performance. While that alone doesn’t tell you anything about the quality of your service , it shows you how busy your live chat is. In other words, whether your visitors are actually using your chat.
Which Operator is handling most chats?
Impressions not only tell you how many chats you got in total, but also how many chats each Operator took care of. That way, you can easily compare the performance of your individual Operators. If you notice significant differences by otherwise constant factors (e.g. Operators have same amount of chat slots, equal product knowledge), you might want to reassess your chat distribution logic .
Why am I not getting a lot of chats?
It might happen that you're not getting as many chats as you’d expected. One reason for this could be that your setup isn’t ideal. Try tweaking it a bit, for example, by adjusting the position or look of your Widget, so it fits your website’s style better. The chat mode you choose can also make a big difference. A proactive mode leads to many more chats than a register form.
Take a look at our video below to see how to make these Widget adjustments.
First Response Time
Compared to other contact channels , live chat is the one where users have the least patience. Oftentimes, it’s just a small question that’s holding them back from completing their purchase. Take too long to reply, and you risk that they’ll look for an answer somewhere else. That’s why the First Response Time (FRT) is widely seen as the most important KPI in live chat.
How fast should I reply to a new chat?
In the Dashboard, we measure the First Response Time as the time in seconds until an Operator responds to an incoming chat. Your goal will be to keep your First Response Time as low as possible. We aim to keep it below 15 seconds.
Keep in mind, however, that this is nothing more than its name suggests. A first response. So you don’t need to solve the issue with that yet. A friendly “ Hi, how may I help you ” does the job. It's simply the “first sign of life” that your visitor gets from you.
How can I boost my First Response Time?
The number of chat slots your Operators have has a negative effect on your First Response Time (more chat slots = higher FRT). Which makes sense; you’re faster to reply if you’re not involved in several other chats already.
Enabling chat macros will speed up your First Response Time as well. While we do love using macros to increase chat efficiency , however, we’re actually not so fond of using them for the greeting. How’s that?
Looking for better customer relationships?
Test Userlike for free and chat with your customers on your website, Facebook Messenger, and Telegram.Read more
With live chat, many visitors aren’t sure if they’ll be talking to a real person or a chatbot . And a canned message might make you sound like a robot – which is never a good idea . While it’s possible to create a naturally sounding macro for the greeting, you’ll risk that recurring visitors notice the redundancy of your first messages. A greeting is a great opportunity to give a more personal, warm touch to a live chat conversation. Use that to your advantage, even if it takes a bit longer.
The best way to boost your First Response Time is to have a team of capable people that can quickly switch between different chats while still ensuring a friendly welcome. Learn how to exercise the necessary skills for that in this post .
The total average of time between responses in seconds. A quick example. Let’s say, your chat session was solved within 4 responses, with respective Response Times of 20, 60, 95, and 25 seconds. Your Response Time will then be 50 seconds ((20+60+95+25)/4)).
Response Times vary for different contact channels . Live chat requires an immediate response, under 2 minutes.
This KPI also doesn’t stand alone. You probably don’t want your visitors to feel you’re trying to kick them out of the chat as fast as possible. Of course, solving a chat interaction fast allows you to move on to the next one, resulting in an overall higher efficiency. But if you make the Response Time your #1 KPI, you risk causing negative effects on your quality – which then affects your customers’ satisfaction with your service.
How can I lower my Response Time?
Setting up macros is a great way to reduce your Response Time. Analyze which questions your customers often ask in the chat (e.g. “Why are you better than your competition?”, “Where can I upgrade my account?” ) and boost your Response Time with smart canned responses.
What’s more, the Response Time usually grows with the number of simultaneous chats your Operators have. So, if you want to lower your Response Time, consider limiting your Operator’s amount of chat slots.
Like with the FRT, it pays off to have well-trained, empowered people in the chat that don’t have to, say, scroll through your help section or check with their superior before they can answer a request.
How happy are your visitors with your chat service? It’s not so easy to measure customer satisfaction , as it’s less straightforward than, say, your First Response Time. That’s why the post-chat survey is so practical.
If you enable the satisfaction rating, your visitors will see a five-star rating right after the chat has ended. They can also comment with more detailed feedback. This can be valuable information to help you improve your service.
Compared to other contact channels, it's quite easy to give feedback with live chat. The reason for this is that you’re asking the customer right after the chat, in the same interface, and with the click of a button. They don’t need to take an extra step.
Which Operator is doing the best work?
Customer satisfaction is a crucial factor to separate the wheat from the chaff in support. Star reviews are helpful tools to find out how well your individual Operators succeeded in making your customers happy. The average score allows you to easily compare their performance.
Got it, 5 stars is great and 1-star ratings suck – right?
It’s not that simple. While star ratings are typically good indicators for service quality, they can also be deceiving. A 5-star answer isn’t always worth 5 stars. Sometimes, the Operator might not know the product well enough and tell the visitor incorrect information, for example, “ Of course you can integrate our product with Asana :) ” – while in fact, it’s not possible. This might satisfy the visitor right after the chat, but if she later finds out that things don’t work the way she was told, you might have to deal with an irritated customer .
Your Operator might also invest too much time in the chat and, for example, listen to the visitor’s whole life story after the issue has already been solved. That ‘free counseling session’ leaves a happy visitor but greatly affects your chat efficiency. So, make sure not to accept positive reviews at face value.
Free Guide to Professional Communication in Customer Support
Download our free guide and master professional customer communication across channels.Get your guide
Negative reviews are the other side of the coin. Even if you’ve done everything right, your visitor might simply not accept your reply and will show her dissatisfaction with a bad rating. That happens quite often, for example, when you don’t offer a feature that they want. And then there’s also Internet trolls that just get joy from creating confusion.
To find out if it’s been an unfair review or if something had gone wrong, we take a closer look at all 1- to 3-star reviews we get. We also check the 5-star reviews on a random basis to make sure the answers our service reps have supplied were all correct. Our support team meets weekly to discuss the findings and prevent mistakes from happening again.
How do I keep my chat team motivated?
Post-service surveys also have another, easily overlooked power: employee motivation. In contrast to other channels like email, you get an immediate response for the service you’ve delivered in the chat. From my own experience of doing chat shifts, I know that a seemingly simple thing such as getting 5 stars and a “ great service :) ! ” feedback can really put a smile on your face – and give you the energy you’ll need for dealing with the tough situations in support .
Another KPI we track at Userlike are Missed Opportunities . These show you how many chats you’ve missed because your chat service hadn’t been available for your visitors. Reasons for this could be either that your Operators were offline, busy, or on away status. So, your visitor clicked the chat button on your website, but weren’t able to start a chat.
Keep in mind that the number of Missed Opportunities depends on your availability. If no Operator is available in the chat, no chat can be started. So if you get a lot of offline messages outside of your service hours, you might consider expanding your chat support during those hours to better satisfy your visitors’ needs.
When do you run out of chat capacity?
Taking a look at the Missed Opportunities helps you get a better feeling how well you’ve staffed your chat. You might find that there are specific times of the day where your Missed Opportunities peak, for example, at the weekends or at night.
How can I reduce my Missed Opportunities?
One way to reduce the number of Missed Opportunities is to staff your chat accordingly. While that might work for very large companies, most businesses don’t have the resources to do so. That’s where Chat Butler comes into play.
We know it can be hard to always be available in the chat. Chat Butler greets your visitors and collects their questions when you’re not in the Chat Panel, while notifying you to come online and take the chat. You don’t have to wait around in the Chat Panel anymore for a new chat, but can come online right when you’re needed, decreasing your Missed Opportunities.
With our Topics function, you can create your own topics to better categorize your chat sessions. They are a way to tag your chats, so everyone knows what they were about without having to read through the whole session.This also allows you to see at first glance how many same-type requests you’re getting.
You can align the topics you set across channels to track whether you’re getting the same kinds of questions in, say, live chat and email, or whether they differ.
“How can topics help me be more efficient?
We assign topics to our chats, so it’s easier for us to take action. Topics we have are “Feature Request”, “Cancel”, “Tech”, “Nonsense” , and more. You can create a topic for any scenario you want. But we recommend to put a cap on your number of topics, so your tracking doesn’t get too small-scaled.
Some examples of how we use topics:
- We mark chats we get from Internet trolls with “Nonsense” , so we don’t waste our time on them when analyzing the 1-star chats.
- We regularly go through the “Feature Request” chats to check if many different users feel there’s one important functionality missing in our software.
- With the “Cancel” topic, we track how many customers chose to reach out to us in the chat to cancel their subscription. Since that’s a critical process for every business, we then also check how well the Operator handled the request.
No KPI is meaningful without a context. As we’ve seen, not even a five-star rating is. Most KPIs are interconnected and depend on other factors, like the number of chat slots. For a holistic overview of what’s working and what isn’t, it pays to look at how different key factors play together.
For a more data-focused customer service setup, also check out our other posts: