Unpacking Call Center Average Handle Time

Average Handle Time, the total time a call center takes, divided by the total number of calls it receives or makes, is never a good thing on its own. Speaking as a manager from an analyst background, if you have access to data including average handle time, then it is a duty to report on the more discrete features of your activities. For example, a call center may indeed take a thousand calls, but how many are really useful to you, or, indeed, the customer?

A particular example offered is that when we look at a proverbial 1,000 calls, only two thirds were of sufficient quality to actually be deemed as a relevant call. Almost 350 calls were made to us for totally different reasons, other than our stated target. Most companies use an ACD on inbound centers, (Automatic Call Distributors), which have a service level defined and built into it, e.g. 85 percent of all calls to be answered within 20 seconds. Duty managers monitoring the queues frequently report that too many calls are held in queue for far too long and that they fail the service level as a result. Obviously, this is a simplistic statement, but there are some important factors to consider even before the call center representative starts to receive inbound calls.

The Value of Automatic Call Distributors

The Automatic Call Distributors detect how many call center representatives are logged into the system and taking calls; talking, whilst it is looking to place calls to available call center representatives (call flows). It may be that you have far too few call center representatives available to handle the incoming traffic, which is a resource pattern issue, or far too many call center representatives in call swing, (waiting for calls), which is also a resource pattern issue. There are other features which frequently facilitate the need to monitor the call queues closely, as follows…

Many duty managers report that the Automatic Call Distributor is recording too many calls in queue and that customers are waiting too long, but the fact remains that many customers are not as patient as they used to be, preferring to hang up and redial in the hope that their call will somehow get advanced to a waiting call center representative, when in reality this is the reverse. The Automatic Call Distributor may indeed be reporting they have 20 calls in queue, but how many of these are people that have hung and redialled? The Automatic Call Distributor queue will show you that it thinks there are 20 people waiting when often you find that sometimes up to half of these people have come back into the same virtual queue again, making it only 10 people, but showing 20 calls! It also artificially explodes your service level failures when you look at the queue report and the end of the day, because the Automatic Call Distributor will record hang ups in queue, but unfortunately we haven’t been able to develop a system whereby we can detect how many of those hang ups have been redialled. There are systems, however, that are designed to cut in and give the customer the option of leaving a message and that they can be contacted later, but that only gives the call center managers one more headache: to service the live calls and then handle the call backs.

The Call Back Conundrum

Once the call has actually been received by a waiting call center representative, what happens then? Remember, the call center representative has recorded this as an inbound call, so is it a useful call to either customer or the call center? This is perhaps where we need to be careful about AHT (Average Handling Time). This one call has come into the call center, but the customer has mistakenly kept this number from a previous experience and thought to use it again, for whatever reason. Only this time the caller is asking for something that the call representative cannot do at this time; it may be a customer service call, or indeed something totally different. Nevertheless it’s a call and the call center representative feels honor bound to try and sell on the back of it. The customer quickly decides that the call is a wasted effort and simply wants to end the conversation and the call center representative tries to record as much information as possible.

This is indicative of a poor call center Key Performance Indicator reporting suite—simply recording how many calls coming in and how many times you sell on them, as well as the time taken to handle these, isn’t understanding the discrete nature of the enterprise. The answer is to decide what constitutes a genuine opportunity from those calls that yield a recordable result and those that cannot be recorded as such. This is where APT (Average Presentation Time), takes on significance. That’s also why we should have inbound call codes. Marketing codes help our call center representatives report on those calls that they can sell to, or help with, and those which they cannot.

When you weed out those calls from the total and calculate the time taken in handling these then that’s when you can really look at genuine productivity. Another important linked feature is call center management, or rather training and development. Far too much emphasis is placed on "turning the customer around," overcoming objections and handling, etc, when often the management should see a genuine "none selling" call for what it is: a misplaced and unproductive call. They should then work vigorously to understand the wider issues of why these non productive calls are coming in. Put all these factors together and this can significantly alter your view on call center productivity as a concept altogether.

Real World Example of a Failed Campaign

Recently we launched a new free phone inbound number for one of our affinity partner. It was a joint campaign to generate data and extend the brand, but all we got initially was loads of calls from customers who actually treated it as something they perceived to be a customer helpline number for that company instead, being of little value to everyone. This then is the all important feature of all the elements that make up a marketing activity, whether inbound or outbound, and I offer the following…

[eventPDF]

• Campaign design (It must be robust and fit for purpose.) You need to know from the outset what you want to achieve from such a campaign or activity.
• Projected forecast of response (This should be set during the trial release period, prior to major campaign launch.)
• Prepare the relevant call scripts for call center representatives to take the customer through the journey, or experience. Far too little importance is placed on this feature and it can significantly alter both Average Handle Time and Average Presentation Time.
• Call Scripts—Decide very quickly through the call scripts and the appropriate result codes, which calls constitute a presentation sale and presentation no sale, whilst recording those calls that cannot be handled appropriately by the call center. That way you get a proper feel for the time taken in actual call center productivity against those calls which are merely "soaking" up resource time and have a wider negative effect on both customer expectation and the morale of the call center representative. Record the time taken to go through a full presentation and use this as a benchmark against the time taken in achieving your presentation sales; often a product may take many tens of minutes to explain to the customer, not to mention overcoming objections in pursuit of closing the call.
• Call Backs—Keep them to an absolute minimum; our standard is no more than 10 percent; remember if call center representatives have to log out of the system in order to deal with these seemingly "nailed on opportunities" then don’t let them set them; it only massively skews the time you think these call center representatives are working, and it is routinely a giant "get out of jail card" for the call center representatives if they are allowed to get away with it. Quickly demand that all call backs are set as a Key Performance Indicator for the call center representative’s performance and this should take care of the issue. Also, that way you don’t have daytime call center representatives setting vast amounts of call backs for the poor later shift guys to deal with, when they are bound to be taking better calls anyway.
• Hang ups—Look at the level of customer hang ups during the opening of a call. Ask why to find out if there was an issue with the call center representative, a controversy or the customer decided they could not be helped and they were not interested in hearing about "other opportunities." Remember, you may indeed turn inquires into opportunities on the basis you can provide the solution for why the customer telephoned in the first place, if not, it’s a bit like trying to sell cheese at a meat counter when you haven’t got any meat to sell—doesn’t work!
• Internal transfers—Keep an eye on internal transfers; are call center representatives allowed to transfer customer calls to other departments and if so does your system keep recording this time; many simply do not, hence the apparent black hole in your reporting when you know call center representatives are working, but the system isn’t recording it because of internal transfers; it also offers another great ‘get out of jail card’ for the call center representatives and again can vastly skew Average Handle Time.
• Average Handle Time—Average handle time is calculated recording the total of calls taken and the total time taken to handle them, but this is nowhere near reflective of your call center operational performance. It’s the time taken in handling those calls you can deal with (Average Presentation Time) and those which you cannot. I have often found the difference in-between average handle time and average presentation time—is many tens of minutes, depending on complexity or service or product offered.
• More importantly; if you have access to all this information; which you should, then this should not only be the KPI’s for the call center representative; it should also be the KPI’s of everyone else; including senior management. You cannot measure one party by one standard and then the other differently, it’s inconsistent and simply means that one set of people are looking for one solution and the other busy dealing with theirs.

Simplifying the Call Calculation

There are 1,000 calls expected, which are being handled by 20 call center representative over 5.5 hrs = 9.0 calls per hour, per call center representative. (110 production hours); based on the assumption your call center representatives are working to standard 80 percent effective time; 5.5 hours out of 7.5 available, taking away breaks, etc. Non productive calls 33 percent = six actual opportunities per hour, per call center representative. You should know your expected sales targets buy this time, say 20 percent; that’s around 1.8 SPH per call center representative, assuming your call flows is steady; which invariably means they won’t be in the real world; those 20 agents should yield you 198 sales in total per day (36 SPH).

Now also here comes the weird bit: time of day, which again is necessary to plan for resourcing reasons. In the UK; I couldn’t comment on the U.S. market at present—customarily we see essentially four peaks of demand on a typical weekday.

9:00 a.m. to 10:00 p.m. Contact rates are fairly steady as people are probably telephoning inbound before either starting work or setting out for work, taking the kids to school, etc.

12:00 p.m. to 1:30 p.m. People tend to telephone when they break for lunch, etc.

4:00 p.m. to 5:00 p.m. People are returning from work, picking the kids up, etc., then it goes quieter for an hour or so; people having their tea, etc.

7:00 p.m. People may be responding as a result of a media campaign.

Keeping people on to take advantage of these flows is a skill in itself and normally the domain of the resource planner, or queue manager; many smaller centers employ a dialler manager to do all three functions, in addition to others.

In summary then, there are many other factors you need to take into account; Average Handle Time is neither a proper benchmark for genuine performance or to observe as a primary driver as it can change massively throughout any given day. The Average Handle Time will certainly be lower during the day and higher in the evening; effective dialler management is all about common sense and perhaps thinking "What would I do—taking into account the many factors that govern the human world?"

There are also other socio-economic and demographical reasons why inbound calls ebb and flow. For example, the weather is often seen as the dialler manager’s friend as these volumes tend to increase when weather is severe; people stay at home and plod around the home, taking care of those mundane chores, hopefully feeling they should give you a call?

All these reasons and many more must be measured and then once you have detected the time taken to handle appropriate calls, only then can you get a handle for a proper Average Handle Time, but perhaps more importantly Average Handle Time doesn’t even give you a reliable figure for how customers feel about your service; all Average Handle Time is doing is telling you how fast are we at receiving calls.

First published on Call Center IQ.