A customer’s time is precious, and so is yours. Let’s get right into everything you need to know about average handle time, how it impacts the customer experience, and how to account for its limitations.
Average handle time (AHT)
Average handle time (AHT) is a metric that is often used as a key performance indicator (KPI) for call centers. It measures the average length of contact for a customer on a call.
Why is average handle time important?
Average handle time is used to assess the efficiency of agents and the customer service organization as a whole. It can be an effective metric for establishing benchmarks or new customer service goals.
66% of customers still typically resolve their issues with a company via telephone, according to the latest Zendesk Customer Experience Trends Report—which makes AHT is an important KPI to track.
66% of customers still resolve issues with a company via telephone
While it’s primarily used to measure the durations of customer phone calls, a company that takes an omnichannel approach to their customer service can use AHT to compare their phone support AHT with other channels (like the duration of live chats).
Here’s what’s accounted for in AHT:
- Total talk time
- Total hold time
- Number of calls handled
What’s a good average handle time?
Average handle times differ depending on a company's approach to the customer experience, the products or services they offer, and the structure of their support organization. Typically, AHT helps a support team accomplish the following:
- minimizing hold times
- optimizing talk times
- increasing the number of calls handled
- improving customer satisfaction
Those goals can indicate an efficient call center. But here’s the thing: lower AHT does not necessarily mean a call center is performing as well as it should. It’s extremely important that agents don't push customers off the phone to decrease their AHT, even if it's part of their responsibility to reduce it. Average handle time shouldn't typically be too high—it should be optimal for the help that customers need and what the contact center is capable of.
How do you calculate average handle time
To calculate average handle time, use this formula:
[Talk + hold + follow up] / calls = AHT (calculated in minutes or seconds)
For example: 150 calls that average out to 3,000 minutes, plus total hold time of 700 minutes, plus follow up time of 500 minutes, divided by the amount of calls. The resulting 28 minutes is the average handle time—way over the industry standard! According to Call Centre Magazine, the industry standard AHT is 6 minutes and 10 seconds.
A similar formula can be used to calculate similar metrics about other channels, like messaging and chat or email.
Many tools can calculate AHT for you. The Erlang Calculator for call centers can be used in a browser window to make quick calculations with some additional variables. This is better suited for one-off calculations and small contact centers. However, if you’re looking for the historical analytical component of AHT, or you recognize the value of collecting big data over the long term, it’s probably best to use software that pulls your support data and creates easy visual charts and dashboards. These dashboards are ultimately better for when you need to share support data across the organization. Realistically, your call center incorporates newer channels. Rather than calculating AHT from each channel and adjusting each time for variables, you can let the analytics software do the hard work.
How do you reduce average handle time in a call center?
If the overall average handle time is higher than it should be, there are a few methods that support managers can employ to bring it down. Here are a few steps to reduce AHT while avoiding the risk of agents rushing through a phone call:
Steps to reduce average handle time
- Make sure agents are trained effectively
- Use self-service resources, like knowledge bases and help articles
- Monitor agent performance
- Record calls to use in ongoing trainings
- Optimize call routing and internal communicationsity for sales teams.
1. Make sure agents are trained effectively
An untrained agent can be a major detriment to AHT. Agents that haven't received proper training may be more susceptible to fumbling through a call or go off on tangents that waste their time and that of the customer. An effective call center thrives when agents are capable of delivering a supportive conversation that's flexible while not being overly responsive to distractions.
2. Use self-service resources, like knowledge bases and help articles
Self-service content can do more for agents than just assist their customers. If a company has invested in a thorough knowledge base, it can be of use to call center employees. Help articles should provide quick access to a list of how-to's on specific issues. Not only does this help train agents, but it also makes them more familiar with common issues and gives them the perspective of a customer trying to solve a problem all on their own.
3. Monitor agent performance
Keep a close eye on the other metrics surrounding AHT. Some of the key metrics for call centers include:
- Average talk time
- Calls missed
- Calls declined
- Transfers accepted
- Average wait time
- Longest wait time
- Abandoned in queue
- Exceeded queue wait
- Average time to answer
- Average hold time
To fully understand the strengths and weaknesses of a contact center and its effect on the customer experience, managers need to know the subtext that comes with those KPI's. There's a good chance that they'll have some effect on AHT. To learn more, read about the 9 key metrics to transform a contact center.
4. Record calls to use in ongoing trainings
Like how a football team watches films of their previous games, agents in call centers should record their calls to review them for ongoing training. Recorded calls make for a great training tool so that agents and their managers can optimize the experience provided by the call center. They can get a good grasp on how customers react after a long hold time, how they engage during the talk time, and the customer experience as a whole. When they better understand those, it can contribute to finding an optimal AHT.
5. Optimize call routing and internal communications
It's crucial for call centers to have their routing processes figure out, as they can contribute AHT and the customer experience. Calls should be routed to the right agent whenever possible - Routing callers to the wrong agent takes up valuable time. A well-designed IVR routing system, or phone tree, will allow callers to select who they need to speak to, so they are connected to the correct agent on the first call.
Agents should also be able to contact each other and collaborate privately within their own workspace. It shouldn't require a third-party app to collaborate with a fellow agent - having a system that can accommodate quick and effective internal communications results in higher customer satisfaction.
Connecting AHT to the customer experience
Customers want efficiency in any support interaction they have—if they have to get in touch with a call center, it's usually because of an inconvenience. AHT can be viewed as an important metric for mitigating that inconvenience and as an indicator of the efficiency that customers want out of an inbound support call.
If a high volume of support calls is having a negative effect on AHT, consider other self-service options to offset those calls. Chatbots and virtual customer assistants can be of assistance for desktop and mobile users, and promoting them via the IVR can potentially reduce the number of support interactions over calls. There are lots of elements that contribute to efficiency, but they can lead to low AHT and better customer experiences.
Average handle time drawbacks: Why you should take the data with a grain of salt
Average handle time remains one of the most popular call center metrics, but the truth is that it can lead to inaccurate insights if not used carefully. Drawing inaccurate conclusions is the last thing any call center leader wants to do, because it impacts their ability to make informed decisions about staffing, training, promotions, and more.
“My strong recommendation would be to use median instead of average,” says Dave Dyson, Community Marketing Specialist at Zendesk and call center veteran.
Remember, this is the formula for average handle time:
[Talk + hold + follow up] / calls = AHT (calculated in minutes or seconds)
To find the median handle time:
- Arrange a set of data points (each interaction’s handle time, in minutes or seconds) from shortest to longest. It’s probably easiest to pick an odd-numbered set, because it’s easy to identify the middle number.
- The middle number is your median.
For example, let’s say your set of five handle time data points is: 1 minute, 1:45 minutes, 4 minutes, 5:05 minutes, 29 minutes. Here, four minutes is the median.
Averages are vulnerable to outliers, and one very long handle time can throw off an average in a way that doesn’t happen with medians, Dyson says. That would be the case in the example data set above, thanks to that 29-minute call.
With handle time, you’re trying to understand the typical amount of time it takes for agents to help a customer. But handle time is especially vulnerable for two reasons:
- Some issues take a long time to solve—the complexity of the issue is likely why the customer reached out to a live agent in the first place.
- The handle time timer is manual. Handle time depends on an agent stopping the timer at the right time.
“If they leave a ticket open while they go to lunch or when they go home for the weekend, the handle time clock keeps ticking,” Dyson explains. “And once the handle time clock is on, it can’t be reset. There will always be some degree of user error in handle time data, and using a median vs. average helps control for that.”
Imagine having a whopping 48-hour handle time in your data set—average compresses all the numbers, whereas the outlier is much more visible when you’re solving for a median.
Averages can still come in handy, such as when it’s important to reconstruct a total, Dyson says. For example, average is accurate and descriptive if you’re analyzing customer spend; you can multiply average customer spend by the number of customers you have and, voila, you get total customer spend.
A good way to use handle time is by mapping data points to specific issues, which helps teams see which areas of the product or service are most time-consuming for agents to support, Dyson says. Multiplying average handle time to get a total amount of time spent on a specific product or service is powerful data for a CX organization.
If you know how much time agents are spending on one area of the business, you can then compare that total to other areas or products. Combined with average agent cost (salary, benefits, infrastructure), it gets even more impactful: you then know how much money it costs to support that area. Knowing this can help team leaders advocate for product or process improvements—the ultimate customer-centric, cross-functional collaboration. Dyson himself did this as a support team leader to help justify improvements to a software integration.
Another important flaw of average handle time: you don’t always want it to go down. What?
Sometimes, driving average handle time down doesn't make sense for the business. For example, agents at Zappo's are on the clock talking about how they're feeling or how their day is going—they're not solving a problem, per se, but they are building connections with customers from the front lines. Magnolia has a similar model, where roughly half of its call center interactions are "experiential." These customers want to share their experiences or just say hello, which is a testament to the powerful connection they feel with the brand.
Furthermore, average handle time can go up in a good way if the business made improvements to its overall CX operation. If the customer self-service, owned community, or AI-supported automations are top-notch, the only time people need a live agent is for help with more challenging tasks. Those interactions typically take longer, and they should. As always, consider AHT not on its own but as one metric of several that paint a complete picture of the support operation.