Data Gathering and Analysis.

The most obvious source of historical call information will be reports from the ACD (Automatic Call Distributor). In terms of how far back to delve into historical reports, its ideal to have two years worth of past history if its available and if its relevant. Less than two years worth may suffice, but won't provide the most accurate tracking of trends and monthly/seasonal patterns that twenty-four months will clearly show.

In terms of what specific information to gather, the two numbers to look for by half-hour are number of calls offered (NCO) and average handle time (AHT). These two sets of information should be available from any ACD, regardless of how basic or sophisticated the system. Most call centers assume the number of calls offered accurately portrays the workload for which it needs to staff. This assumption is valid as long as "all calls are getting in the door" and that none are blocked at the network level by an insufficient number of incoming telephone lines. It's always a good idea to validate this assumption by requesting periodic "busy studies" from local and long distance carriers.

Offered Versus First attempt Calls

In situations where there is a high demand and not enough incoming lines, the number of calls offered as reported by the ACD will only show the ones that "got in the door." (In some ACD reports, it may be necessary to add together the number of calls handled and the number of calls abandoned to calculate the actual number of calls offered.)

A true traffic analysis should also take into account the blocked calls and the retry behavior of callers to determine a true "first attempt" call load.

Above picture illustrates a situation where not enough incoming lines were available and a significant portion of the calls was blocked. The ACD shows there were 1000 calls offered between 9:00 - 10:00 am. But an investigation with the local carrier shows that 22 percent of the calls attempting to enter the ACD found no incoming line available. Those callers heard a busy signal and were forced to retry their call. Assuming that half the caller might retry during that hour, a true first attempt workload can be calculated.

To arrive at the first step workload, one could work backwards from the number of calls connected (answered and recorded by the ACD as "offered" to the agents) and the blockage percentage to find the number actually offered to the front door of the ACD. If 1000 calls made it into ACD, and 22 percent were blocked that means that the 1000 calls represent only 78% of the call attempts.

Dividing 1000 calls by .78 equates to the actual number of calls that were sent to the ACD, 1282 calls. Therefore, 282 calls were blocked because there were not enough trunks to the call center's ACD, and these callers heard a busy signal.

These callers then made a decision to either call back within the hour or not. And of those calls that were blocked, If half the callers retry, then 141 calls of the 1282 attempts would be retry calls. In a blockage situation, the calls "Knocking at the ACD door" are made up of the first attempts plus retries, and in this case if retires represent 141 calls, the remainder (1142 calls) are the true first attempt calls, as illustrated in below picture.

The Formula for calculating all attempts (sometimes referred to as "offered traffic" in a telecommunications traffic engineering scenarios, but not to be confused with number of calls offered (NCO) as recorded by the ACD) and the calculations associated with the examples are (where C-Carried ACD Calls, and B-Blockage Percentage):

Next the formula for calculating first attempt calls, along with the calculations for this scenario, is as follows (where C-Carried (ACD) calls, B-Blockage Percentage and R-Retry Rate).

In this scenario, if the number of calls offered shown on the ACD report was used as input data for forecasting, 1000 calls would have been used when the actual first attempt number of calls was 1141. Call workload would have been significantly understated, and resulting staffing levels inadequate to handle the demand. Therefore, when significant blockage is evident on the incoming telephone facilities, it is critical to determine what the true first attempt call load was before becoming the forecasting process. Of course, if no more trunks are added to prevent the blockage and busy signals, these calls may still not reach the agents. In many cases, the truck blockage is a result of long queue times, and providing adequate staff to handle the calls will shorten the time the trunks are utilized for each call, and may result in more calls actually reaching the agents without adding trunks.

Thanks,

With Love and Sincerity,

Contact Center Workforce Management and Quality Optimization Specialist