Definitions
Below is a glossary of definitions related to performance metrics. These definitions are assumed on the site (where they are also stated) as well as in feature articles. Should your own definition vary from those listed below, please state so in your post or article and provide the definition that you use. Definitions, of course, vary by telephony platform and, where this is the case, we try to state so. We are always amenable to updating our "official" definitions should a majority of users lobby for their update.
Telephony Metrics:
The percent of time an
agent spends engaged in work activity, as
opposed to waiting for a call. Generally, (Talk
Time + Hold Time + After Call Work) / (Talk Time
+ Hold Time +After
Call Work + Idle Time). In other words, what
percent of the agents day was spent working -
either speaking with the customer or engaged in
after call work, as opposed to either waiting
for an incoming or call or to make an outgoing
call. Please note that one some systems, such as
Symposium, Talk Time is inclusive of Hold Time.
The percent of time an
agent spends peaking the phone with customers.
Generally, (Talk Time + Hold Time ) / (Talk Time +
Hold Time + After Call
Work + Idle Time). In other words, what percent
of the agents day was spent on the phone, as
opposed to either 1.) Performing After Call Work
in a state of Not Ready 2.) Waiting for an
incoming or call or to make an outgoing call. As
you can see from the above, the primary
difference between Utilization and Occupancy is
that Utilization is used to calculate After Call
Work in terms of time the agent could have been
available to handle customers calls. Calculated
at Agent Level. Please note that one some
systems, such as Symposium, Talk Time is
inclusive of Hold Time.
The percentage of calls answered within the timeframe (or threshold) as defined by the company. Basically, the percentage of calls that were answered within what the company regards as having met the companies customer service goals. Calculated at Skill Level. In Nortel, the most common formulation ((Callsanswered+Callsabandoned)-(Callsansweredafterthreshold+Callsabandonedafterthreshold))/(Callsanswered+Callsabandoned). Thus, the calculation excludes those calls that were either answered within the target time and those for which the caller did wait enough time to meet the centers abandoned threshold. Callsoffered can be substituted for (Callsanswered+Callsabandoned), but this can cause metrics to lack transparency as it will include calls that were disconnected. In other systems, such as Avaya, this field is summarized. In Avaya, it is summarized as "ACCEPTABLE". Thus making the calculation ACCEPTABLE/CALLSOFFERED. Top
The total delay time (time that calls are waiting in queue) divided by the number of calls answered for the same period or interval. Often times synonymous with Average Answer Delay, although this can be misleading as both include calls that were answered immediately. For a more accurate assessment of caller wait time, see Average Time in Queue (this is the average wait time for customers who experienced some measure of delay). Calculated at Skill Level. Top
The average wait time for customers who experienced some measure of delay. On other words: (Answer Delay/Calls Answered) where Answer Delay >0. Calculated at Skill Level. Top
The total amount of time it takes a phone agent to handle a call. Inclusive of talk time, hold time and after call work (ACW or 'Wrap Up Time'). Generally expressed as a per Call metric (Talk Time +Hold Time+ ACW)/(Calls Answered). Top
After Call Work (ACW)/Wrap Up Time
The amount of time an agent spend in a state of "Not Ready" following a call. This time is used to complete work the agent was not able to complete for any reason during the course of the call itself. Top
This is often more of a general description than a definition, unless otherwise indicated. When used to refer specifically to Nortel's Symposium definition it will appear as it does in their Sybase tables, as one word: Notreadytime. When used generally, it refers to a state in which the agent is unavailable, for whatever reason, to make or receive calls. Unlike ACW, Notreadytime is an agent level metric and not pegged to a skill set. Top
Service Thresholds are the call centers benchmarks for providing service within a set timeframe (or threshold) and can vary by skill set. Most commonly notated in reporting as the y in x/y. For example, in a Service Level target of 80/20, the 20 indicates 20 seconds, this being the threshold, by which the center aims to answer 80% of it's calls. 90/10 meaning 90% answered within 10 seconds, etc.... The most common thresholds are Call Answer Time and Call Abandoned Time. Top
The number of call answered divided by either the number offered (answered/offered) or the number of calls answered divided by sum of calls answered plus calls abandoned (Answered/(Answered+Abandoned)). Which calculation used depends on whether or not your offered filed includes calls that were disconnected. As with Service Level and Abandoned Rate, it is important to maintain consistency. We prefer to use Answered plus Abandoned in our numerator and feature a separate metrics for disconnects as it avoids the inevitable question of "why don't number of calls answered plus the number abandoned add up to the number offered?". Top
The number of call abandoned divided by either the number offered (abandoned /offered) or the number of calls abandoned divided by sum of calls abandoned plus calls answered (Abandoned /(Answered+Abandoned)). Which calculation used depends on whether or not your offered field includes calls that were disconnected. As with Service Level and Answer Rate, it is important to maintain consistency. A common alternative is to use Calls Abandoned After Threshold in the numerator. While using the latter method gives a clearer picture of performance, it can also mask other issues related to dropped or abandoned calls. Top
The average time in queue for a customer prior to reaching an agent. Average Answer Delay usually includes calls that were answered immediately, so it should not be confused with Average Time in Queue (ATQ), which includes only calls that experienced some delay prior to being connected to an agent. Generally, the total answer delay divided by the number of calls answered. The metric tends to become less meaningful in direct proportion to the specificity of the interval over which it is measured. For example, measure over a specific staffing interval (the smaller the better) for a specific DOW (Weekday). Generically, SELECT SUM(AverageAnswerDelay)/Sum(Callsanswered) FROM X WHERE Interval BETWEEN 1600 and 1800 and DOW = Thursday. Top
The longest wait time a caller experienced prior to being connected to an agent for a specific interval. The metric, like Average Answer Delay, tends to become less meaningful in direct proportion to the specificity of the interval over which it is measured. For example, measure over a specific staffing interval (the smaller the better) for a specific DOW (Weekday). Generically, SELECT SUM(MaxAnswerDelay)/Sum(Callsanswered) FROM X WHERE Interval BETWEEN 1600 and 1800 and DOW = Thursday. Top
The average time in queue for a customer prior to hanging up. May or may not be set to calculate using Abandoned Threshold. Generally, the total Abandoned Delay divided by the number of calls abandoned. The metric, like Average Answer Delay, tends to become less meaningful in direct proportion to the specificity of the interval over which it is measured. For example, measure over a specific staffing interval (the smaller the better) for a specific DOW (Weekday). Generically, SELECT SUM(AverageAbandonedDelay)/Sum(Callsanswered) FROM X WHERE Interval BETWEEN 1600 and 1800 and DOW = Thursday. Top
The longest wait time a caller experienced prior to disconnecting for a specific interval. The metric, like Average Answer Delay, tends to become less meaningful in direct proportion to the specificity of the interval over which it is measured. For example, measure over a specific staffing interval (the smaller the better) for a specific DOW (Weekday). Generically, SELECT SUM(MaxAbandonedDelay)/Sum(Callsanswered) FROM X WHERE Interval BETWEEN 1600 and 1800 and DOW = Thursday. Top
The average number of calls answered per hour. The "hour" can be calculated as login time or as productive time. In fact, there are any number for calculations for this, all of which hinge on the how you define the numerator (hour). This may sound like a non-definition, but most centers (and different projects or campaigns within a center) tend to have their own calculation. Top
The number of call expected for a given interval divided by the number of calls actually presented (offered) for same interval. A common calculation is to do this by either a staffing interval, by DOW, or both. Generically, SELECT Callsoffered FROM X WHERE DATE Between Sysdate-x and Sysdate-1 and DOW = Wednesday. 'x' being the number of weeks you want to use for your calculation. Top
The ratio of work time to login time. Generally, (Talk time+ Wrap Time + Available) divided by login time. As with many metrics, this is one people can get carried away with in terms of 'customizing' (I.e. developing a unique calculation that factors in elements unique to either your center or - more dangerously - your performance goals. While it is certainly best practice to tailor metrics to your needs and goals, this is often done at the cost of transparency and the ability to evaluate other needs. Top
N is the number of servers
PW is the probability that a customer has to wait for service
Yes - we are going to flesh this one out a bit. We are hoping to have an online calculator for B and C available the week of December 3, 2007. Please do stay tuned.
