The number of concurrent chat sessions an Agent can manage is often a source of great personal pride; as well as a great opportunity to increase Agent productivity. But are we basing our decision on maximum concurrent chat sessions purely on Agent enthusiasm at the expense of customer experience?
The recent emergence of web chats as a core contact centre function has introduced a multitude of new and exciting challenges for contact centre management to solve; far too many to discuss in a single article. So for now, I’ll focus on just one question:
How many concurrent chat sessions does it take to destroy a positive customer experience?
The absolute answer will vary from one contact centre to another, but I hope the following example helps to highlight the considerations that need to be made when determining your site’s concurrent chat session limit.
Anatomy of a Single Chat Session
While phone calls demand the attention of both customer and Agent throughout the entire interaction, the participants of a webchat take turns in being engaged within the interaction. While one participant is reading, researching, and responding, the other is waiting for a reply to their last post. Consider the following representation of a chat session.
When the Agent is engaged they are reading the customer’s post and posting a response. During this time, the customer is waiting for the Agent’s response. When the customer is engaged, the customer is reading the Agent’s post, and posting a response, while the Agent awaits their response.
The period of customer engagement prior to the Agent’s first period of engagement, and post the Agent’s last period of engagement, are excluded from the handle time of the chat session. The customer is engaged at this point, but the Agent is not waiting for a reply. On this basis, the Agent Handle Time for this chat session is 240 seconds.
During this example chat session, the customer has waited a total of 110 seconds for replies to their posts; an average of 16 seconds between each post they have made. The Agent has waited a total of 130 seconds for the customer to reply; an average of 22 seconds between each post.
To keep a customer engaged, it is paramount the time between responses is minimised. The greater the time between responses, the greater the chance the customer will become distracted. And a distracted customer is more likely to change their mind about a sale, lengthen their response times (respond during ad breaks, after getting a snack, etc.), or abandoning the chat session before resolution.
Two Parallel Chat Sessions
The introduction of a second parallel chat session offers the opportunity for Agents to be engaged in response to one customer’s post while waiting for a response from a second customer – brilliant! The only complication is the emergence of periods where an Agent hasn’t finished with a response to one customer when the second customer has already completed their post. I’ve highlighted these periods of Additional Customer Wait time in green in the below representation of parallel chat sessions.
With two chat sessions running in parallel, the Agent’s average time waiting for a customer’s response has been reduced from 22 seconds down to just 3 seconds. However, the customers’ average wait time between responses has increased from 16 seconds up to 19 seconds. The Average Handle Time per chat session has also increased from 240 seconds up to 260 seconds. Our Agent is definitely more productive, but customers are experiencing longer wait periods between responses, and the overall duration of the chat session has increased. With a huge reduction in Agent wait time and only a minor increase in customer wait time, operating with 2 concurrent chat sessions looks like a sound solution in our example contact centre. But what would happen if we introduced a third parallel chat session?
Three Parallel Chat Sessions
Adding a third concurrent chat session has reduced Agent wait time between responses to 0 seconds! However, customers are now waiting an average of 23 seconds between posts for a reply, and their chat sessions are averaging 293 seconds in duration.
When ‘wait-times’ between interactions is approaching 0 seconds, Agents are cycling without break from one response to another. At this point, no further productivity gains can be made through leverage of Agent wait time between chat interactions – because there is none. It is also hard to imagine Agents being able to sustain this rate of work for more than a couple of minutes without compromise of quality or Agent inflation of handle times to accommodate the need for short breaks between responses.
Advantages in Pursuit of Service Levels
Further benefits from extra parallel chat sessions could be perceived from a Staff Requirement to Meet Service Level perspective, but these benefits should be pursued with caution. You may be able to initially answer more chat sessions within targeted service levels but at the detriment of inter-chat service levels. As seen with this example, inter-chat wait times have already extended from 16 to 23 seconds. Introducing a fourth parallel chat session would push this out to 34 seconds!
If customer wait times for the initial answering of the interaction are important to your contact centre for maintaining customer satisfaction and sales propensity, they should also remain important during the interaction. It would be easy to argue that abandonment during an interaction causes more damage than abandonment before the conversation has even begun.
Finding Your Equilibrium
I am unaware of any silver bullet formula that will determine your contact centre’s ideal maximum number of concurrent chat sessions. It would be easy to state the ratio between customer and Agent chat engagement can be used as a rule of thumb. If Agent engagement is approximately 50% of an average chat session, then 2 concurrent chat sessions should be the maximum. But the duration of each interaction within a chat session will significantly influence the time a customer will need to wait for a reply while an Agent is responding to another customer.
I believe the most reliable means to determine your concurrent chat limit will be through careful observation, and customer feedback. If Agents are cycling from one response to another without break, your limit has been exceeded. If customers are complaining about response times, or they are increasingly abandoning mid chat sessions, your limit has been exceeded.
Concurrent chat sessions offer a golden opportunity to increase the productivity of contact centre Agents. However, it’s one thing to want three ice creams on a hot summer’s day, but can you eat them without making a mess of them all?
I would love to hear about your web chat experiences. Have you found your equilibrium, or are you still searching for the answer? And remember, while your contact centre must decide for itself what the right number of concurrent chats is, we’re here to help you find the answer!