ICMI is part of the Informa Tech Division of Informa PLC

This site is operated by a business or businesses owned by Informa PLC and all copyright resides with them. Informa PLC's registered office is 5 Howick Place, London SW1P 1WG. Registered in England and Wales. Number 8860726.


The Metrics of Contact Center Productivity

Cost per contact is the most important measure of efficiency in the contact center. It is an easily derived metric that tells you how much you are paying for one unit of work (one contact). If two contact centers perform the exact same service, yet one has a cost of $7.00 per contact, and the other has a cost of $5.00 per contact, the latter contact center is clearly more efficient.

But how can we manage cost per contact, and more specifically, what are the underlying drivers of cost per contact? Many productivity factors drive cost, and one or more metrics can measure each of these productivity factors. These are the metrics of contact center productivity.

Cost per Contact Defined

Cost per contact is the total annual operating expense of a contact center divided by the annual inbound contact volume of the contact center. Operating expense includes all employee salaries, overtime pay, benefits, and incentive compensation, plus all contractor, facilities, telecom, desktop computing, software licensing, training, travel, office supplies, and miscellaneous expenses. Contact volume includes inbound contacts from all sources: voice, email, web, chat, fax, and walk-in.

Contact centers are labor intensive. Approximately 70% of all contact center costs are personnel related – salaries and benefits for those who work in the contact center. Some contact centers will simplify the cost per contact metric by including only personnel costs in the calculation. This avoids the sometimes-tedious task of determining all of the non-personnel costs.

We can further segment cost per contact into 1) cost per agent-assisted contact, and 2) cost per contact for all contacts, including IVR and AI (artificial intelligence). Let’s say, for example, that a regional bank handles one million agent-assisted voice calls per year and spends $6 million annually on its contact center. The agent-assisted cost per contact is simply $6,000,000 in operating expense ÷ 1,000,000 calls = $6 per agent-assisted contact. Now, let’s assume that another 1 million contacts are resolved each year in the banks IVR. In other words, 50% of all inbound contacts are resolved in the IVR (a 50% IVR containment rate is about average in retail banking). These IVR contained calls have no marginal cost, so the total cost per contact, for all contacts including the IVR, would be $6,000,000 in operating expense ÷ (1,000,000 voice calls + 1,000,000 IVR resolved calls) = $3 per contact.

The Drivers of Contact Center Productivity

Before we get into the specific metrics of contact center productivity, let’s start by defining the key factors that drive cost per contact. These factors can be roughly divided into four categories:

  • Agent productivity
  • Channel mix
  • Geography
  • Root cause elimination of contacts

Agent productivity is what most people think of when it comes to making the contact center more efficient. However, it turns out that optimizing agent productivity, by itself, often does very little to move the needle on cost. By contrast, optimizing channel mix can sometimes have the effect of reducing cost per contact by 50% or more.

Channel mix is simply the percentage of inbound contacts that arrive at the contact center through various channels: voice, chat, email, web, IVR, AI, and walk-up. As shown in the example above, deflecting contacts into the IVR can have a significant impact on total cost per contact! This is why so many contact centers devote substantial resources to optimizing their IVR menus.

Wage rates drive cost in a labor-intensive function. As such, large contact centers tend to be located in geographies that have lower wage rates. From a cost perspective, it makes more sense to find a contact center in Des Moines, Iowa than in New York City; or in Manila than in San Francisco.

Finally, contact centers are increasingly taking a page from IT Service Management, and adopting the principles of Problem Management, an ITIL discipline that focuses on the root cause elimination of contacts. Some contact centers have been successful at eliminating up to 30% of their annual contact volume through effective root cause analysis.

Let’s examine each of these productivity drivers in more detail.

Agent Productivity Metrics

The primary metrics of agent productivity are agent utilization, average handle time (AHT), and contacts per agent per month.

Agent utilization is simply the ratio of work produced divided by work capacity. So, for example, if an agent is on customer calls for five hours out of an eight-hour shift, the utilization for that agent that day would be 62.5% (5 hours of work produced ÷ 8 hours of work capacity).

The formula for determining agent utilization is somewhat complicated. It factors in the hours in a workday, breaks, vacation and sick days, training time and a few other variables. But there is an easy way to approximate agent utilization without going to so much trouble. Here is the simplified formula:

Agent Utilization Formula

Let’s say, for example, that the agents in a particular contact center desk handle an average of 375 inbound contacts per month at an average handle time of 10 minutes per contact and make 225 outbound contacts per month at an average handle time of 5 minutes per contact. Additionally, these agents work an average of 21.5 days per month, and their workday is 7.5 hours after subtracting lunch and break times. The simplified utilization formula above would work out to the following:

Agent Utilization Formula, Solved

Once again, this is not a perfect measure of agent utilization, but it is quick and easy and gets you within 5% of the true agent utilization figure.

When utilization is low, the cost per agent-assisted contact will be higher. This could be an indication that the contact center is over-staffed, that schedule adherence is low, or that workforce scheduling could be improved. Conversely, when agent utilization is high, cost per agent-assisted contact will be lower.

Average Handle Time (AHT) is the average time that an agent spends on an inbound contact, including talk time, chat time, wrap time, and after call or after chat work time (ACW). For non-live contacts, such as email and web submitted inquiries, the contact handle time is the average time that an agent spends working on a contact before escalating or resolving the contact.

A contact is the basic unit of work in a contact center. Average handle time, therefore, represents the amount of labor required to complete one unit of work. Additionally, it is an indirect measure of contact complexity. As an example, a balance inquiry to a credit card contact center might have a handle time of two or three minutes. By contrast, a call to an insurance company to purchase an auto insurance policy might have a handle time of 20 minutes or more.

For obvious reasons, cost per contact increases as handle time increases. As such, many contact centers focus on handle time, and specifically handle time reduction, as a means to reduce cost per contact. AHT can be reduced in a number of ways including agent training, scripting, and certain technologies such as screen pops, CRM, and knowledge management. In some of the industry’s largest contact centers, even a handle time reduction of 10 or 15 seconds, when multiplied across hundreds of millions or even billions of contacts per year, can yield huge savings!

Contacts per Agent per Month is the average monthly inbound contact volume divided by the average Full-Time Equivalent (FTE) agent headcount in the contact center. Contact volume includes contacts from all agent-assisted channels: voice, chat, email, web submitted inquiries, walk-in, etc. Agent headcount is the average FTE number of employees and contractors handling customer contacts.

Contacts per Agent per Month is an important indicator of agent productivity. A low number could indicate low agent utilization, poor scheduling efficiency or schedule adherence, or a higher-than-average handle time. Conversely, a high number of inbound contacts per agent per month may indicate high agent utilization, good scheduling efficiency and schedule adherence, or lower-than-average handle time.

Channel Mix Productivity Metrics

Channel mix in the contact center is rapidly evolving (Figure 1 below) and is considered one of the industry’s megatrends. In 2008 voice calls represented almost 80% of all contact volume. Today, voice accounts for just over 50% of incoming contact volume.

There are two key drivers behind this trend. One is economic, and the second is demographic. The key economic driver is that voice, which has dominated the industry for decades, is one of the most expensive channels, while chat, email, web, and especially IVR and AI are all lower cost than voice.

Figure 1, Evolving Channel Mix

IVR and AI contacts are lower cost than voice contacts for obvious reasons. But it’s not so obvious why chat, email, and web submitted contacts are lower cost than voice. Since voice is a real-time channel, the Erlang distribution of incoming calls tends to be very spiky. This, in turn, makes it challenging to staff the voice channel efficiently. The voice channel is perpetually either understaffed or over-staffed! Chat is also a real-time channel, but an effective chat agent can often handle two or more concurrent sessions, thereby making chat more efficient than voice. In fact, MetricNet’s benchmarking data shows that the average contact resolved in chat costs just 76% as much as a contact resolved in the voice channel.

Email and web submitted contacts tend to be lower cost than voice contacts because these channels do not have the same real-time urgency as voice and are therefore less spiky than voice and chat. This makes it easier to staff email and web channels, which leads to better agent utilization and lower costs. On average, the cost of an email or web-submitted contact is about 81% of the cost of a voice contact.

What about the demographic drivers behind channel mix? If you look at a spectrum of customers – from those who are just entering the workforce, to those nearing retirement – younger customers tend to prefer self-help (IVR and AI) or the indirect channels of chat and email, while those who have been in the workforce for some time rely much more heavily on live voice as their go-to contact center channel. This trend will continue as newer workers enter the workforce.

What is the right channel mix, and how does a contact center achieve that state? The answer depends upon the unique nature of the services provided by your contact center. It would be tempting to suggest that the ultimate channel mix is to drive all contacts into the IVR, thereby eliminating all costs associated with voice, chat, email, and web support. But this is neither realistic nor wise. Many contacts simply cannot be resolved in the IVR, and many customers would prefer to interact with a live agent rather than an IVR.

There are, however, better end states for channel mix, and less optimal end states for channel mix. The following principles guide the contact centers that come closest to truly optimizing their incoming channel mix. First, they ensure that customers have a great experience regardless of which channel they choose. Offering chat might sound good, but a poorly implemented chat channel can harm both the customer and the reputation of the contact center. Secondly, they never force a customer into a particular channel. Instead, they give them incentives to use lower cost channels. Two examples come to mind here.

Every voice or chat session should be viewed as an opportunity to educate the customer. If you simply resolve a caller’s issue without providing a re-usable solution, they will contact you again and again with the same issue. This is akin to the “give them a fish vs. teach them how to fish” allegory. If you can teach a caller how to resolve some of their own issues quickly, you can reduce overall contact volume while simultaneously empowering the customer. This could be as simple as pushing a knowledge article to the caller or walking the caller through a co-browsing session to perform a transaction on your website. A second common practice is to play a wait time message when the voice channel is experiencing heavy volumes. Sharing simple information such as the expected wait time and a URL for chat, email, or self-help support can encourage trials and ultimately adoption of a new (lower cost) channel by the customer.

Finally, they minimize costs, consistent with principles one and two. In other words, they are proactive about driving contacts into lower cost channels, but they never override the will of the caller or sacrifice the customer experience in doing so.

Channel mix will continue to evolve for many years to come. Voice is decreasing as a percentage of all incoming contacts, while chat, web, IVR, and AI are all on the rise. There are both economic and customer experience benefits to be had by proactively managing this evolution. Customers can and should be encouraged to use lower cost channels, but never at the expense of the customer experience.

The Productivity Metrics of Geography

Cost per Contact is a function of geography due to wage rate differentials. For a labor-intensive function, any reduction in wage rates will have the effect of reducing the cost of agent-assisted contacts. To follow up on a prior example, if entry-level contact center agents in New York earn $45,000 per year, while entry-level agents in Des Moines earn $30,000 per year, the agent-assisted cost per contact will decrease in proportion to the wage rate difference if the contact center is moved from New York to Des Moines.

Take the case of an insurance claims contact center in New York that is spending $30 per inbound contact. The same contact center would spend just $20 per contact in Des Moines (($30k salary in Des Moines ÷ $45k salary in New York) X $30 per inbound contact in New York) = $20 per inbound contact in Des Moines. This simple calculation assumes that all non-personnel costs will also decrease in proportion to the wage rate differential, which is, roughly speaking, a valid assumption.

The proliferation of offshore contact centers since the mid-1990s is a direct reflection of the wage rate differential between higher wage countries in Europe and North America vs. lower-wage countries in southeast Asia and elsewhere. To continue with our example, and just by way of illustration, we could reduce costs still further by moving our claims contact center from Des Moines to Manila. The average contact center agent in Manila makes about $5,000 per year. So, our $20 cost per contact in Des Moines would become just $3.33 per contact in Manila (($5k salary in Manila ÷ $30k salary in Des Moines) X $20 per inbound contact in Des Moines) = $3.33 per inbound contact in Manila.

The Productivity Metrics of Root Cause Analysis

In information technology, there is a practice called Problem Management. The idea behind problem management is to eliminate the underlying drivers of incidents that generate support calls to the service desk. Some companies have been successful at reducing their ticket volumes by 50% or more through problem management.

By the same token, contact centers that engage in root cause analysis endeavor to identify the key drivers of contact volume. By focusing on the underlying drivers of contacts, it is often possible to reduce volume significantly. For example, one financial services company sought to reduce the number of agent-assisted password reset calls by adopting a bio-recognition password tool for voice and fingerprints. Within a year the volume of password reset requests decreased by almost a third, and overall volume was reduced by approximately 2.5%. This may seem like a modest improvement, but for this particular contact center, which handles more than a billion consumer calls a year, the reduction in agent-assisted password reset calls alone saved them tens of millions of dollars.

The productivity metric for root cause analysis is the decrease in contacts per customer, month-over-month and year-over-year. One major gas and electric utility redesigned their monthly billing statement to make it easier for their customers to understand. Additionally, they began pushing text and voice messages to customers who were affected by outages. The combined impact of these two actions reduced the annual contact volume per customer from 1.26 calls per year to 1.14 calls per year. For this particular utility, who had several million customers, the net effect was a decrease of almost a million calls per year and a savings of more than $7 million a year!

Maximizing Contact Center Productivity

As a 30-year veteran of this industry, I have seen contact centers struggle more often than not with cost containment. Many times, these contact centers do not understand the connection between productivity and cost, or they view productivity too narrowly, strictly in terms of agent efficiency and utilization. By viewing productivity more broadly, as any factor that impacts cost, it is possible to open up new avenues of productivity improvement. Some of these productivity factors – including channel mix, geographic location, and root cause analysis – have the potential to reduce the overall cost of your contact center dramatically.

Topics: Analytics And Benchmarking, Contact Channels, ROI, return on investment