Determing the size (number of telephone lines/ports)
of an IVR system, prior to installation, is always a difficult
process.
As an example, Telephone Banking is one area
where customers do not generally spread the calls evenly throughout
the day or week. Depending on the customer base, the reasons
for the calls may vary:
- Those receiving electronically deposited payrolls may
be repeatedly checking their account balance to see if the
vital funds have arrived.
- Those with young families may wish to check early in
the morning before dropping the children off at school.
- Workers may check their account details on arriving at
the office or just prior to lunch hours.
The result may be one day in the week (or
one moment in a day) where everybody is trying to get access
to the IVR system at the same time. If customers are unable
to use the service, when and where they need it, they may
return to less cost-effective service delivery channels, such
as physical Branches, Automatic Teller Machines, or making
calls to staffed offices.
However, if you are able to make an assessment
of the anticipated call volumes, their distribution during
the day, and average call durations then you can take advantage
of a branch of mathematics developed by a pioneer in this
area, A.K. Erlang in 1917, from which a formula bearing his
name was developed.
As it would be uneconomical to design a system
such that 100% of received calls were successful, industry
standards dictate that it is acceptable if 99 out of 100 are
answered on the first call (ie. a 1% call blocking factor).
With this assumption you can then use one of Erlang's sets
of tables to estimate the number of telephone lines/ports
required.
Using this information the following simplified
table can act as a guide for planning:
| LINES |
ERLANGS |
PEAK
HOUR CALLS |
| 45
sec avg |
60
sec avg |
75
sec avg |
| 4 |
0.87 |
69 |
82 |
42 |
| 8 |
3.13 |
250 |
187 |
150 |
| 12 |
5.88 |
470 |
352 |
282 |
| 16 |
8.87 |
709 |
532 |
425 |
| 20 |
12 |
960 |
720 |
576 |
| 24 |
15 |
1200 |
900 |
720 |
| 28 |
19 |
1520 |
1140 |
912 |
| 32 |
22 |
1760 |
1320 |
1056 |
| 36 |
26 |
2080 |
1560 |
1248 |
| 40 |
29 |
2320 |
1740 |
1392 |
| 44 |
33 |
2640 |
1980 |
1584 |
| 48 |
36 |
2880 |
2160 |
1728 |
| 52 |
40 |
3200 |
2400 |
1920 |
| 56 |
43 |
3440 |
2580 |
2064 |
| 60 |
47 |
3760 |
2820 |
2256 |
| 64 |
51 |
4080 |
3060 |
2448 |
IVR Capacity Reference Table for 1%
Call Blocking
Once you have a Swift IVR system in operation, however, it
provides a quantitative measures of the system's utilisation.
The system's standard reports will show vital
information like the number of times all lines were in use,
average and peak call duration and hourly usage graphs. The
system also computes an Erlang value for its peak hour calls
and indicates the effective number of lines required to service
the call volumes experienced.
We have found that, in practice, if all lines
are in use for more than a few occasions each day, the system
is at capacity and further call volume growth will be affected,
so you have good reason to monitor a system’s operation.
Sometimes adding extra lines/ports is not
necessary, all that may be required is to reduce the average
length of a call by a few seconds. This can be achieved by
careful design of the structure of menus that customers hear
so that commonly used services are played first, removing
verbose prompts and lengthy interactions, or by the overlapping
of host computer system access times with other activities.
However, we do try to be very aware of these issues when a
new system is first set up so there may be limited scope for
subsequent change in this area.
Swift Call would be happy to assist you in
the evaluation of your needs in this area. Our experience,
and insights into caller behaviour, can help you make the
right assessment.
|
