
PREDICTIVE DIALING ![]()
Predictive
dialing offers the greatest return on investment from automated dialing systems.
The underlying technology is complex and is almost always described only superficially
by most vendors. Because of the high levels of automated dialing, what really
takes place is not easy to determine. This may be the reason that while every
dialer vendor claims to have the best predictive dialer, independent and objective
evaluations are not available.
With predictive
dialing, we are able to achieve very high rates of operator productivity
by reducing the time between each collection call. The dialer has traditionally
been able to do this by setting up a group of collectors in a "pool"
and then placing more calls than there are agents, knowing that there will be
a percentage of calls such as the busy's and no answers that do not have to
be transferred to an agent.
These unconnected calls are documented in the system, and management is able
to act on the results just as they would had the agent placed the call themselves.
But because the entire process is automated, there is no effort expending in
dialing the phone and no time is wasted on managing or documenting the unconnected
calls. With the appropriate "pacing" technology, the dialer is able
to predict the availability of agents and deliver connected calls with as little
idle time as possible between each call.
The following benefits are realized with well-managed predictive dialing operations :
There are some disadvantages of using a predictive dialer in a collection environment. These are the following :
When predictive dialing
is applicable, you usually work with different campaigns that are set up to
target specific groups or types of accounts. With I-Tel, it is also possible
to set up "agentless" campaigns which may be started outside normal
office hours or for special types of accounts. Messages can be played when calls
are answered or answering machines are detected.
ABANDONED CALLS
Any
time you dial more than one call per agent, there is always the chance that
you will get more than just one live call per available agent. Since
an agent can only handle one live call at a time, it means that any other live
call(s) cannot be dealt with immediately by an agent. Different dialers use
various strategies to minimize and draw attention away from such calls. Recent
legislation has highlighted the area of abandoned or nuisance calls, and more
people are aware of the issues. However, for the average user, there are still
more questions than answers. A well-designed dialer will launch calls such that
the expected number of answered calls equates to the number of agents available,
given that some numbers can be expected to be busies, no answers etc. But statistically,
any time you dial out on more trunks than there are agents, you run the risk
of nuisance calls. If you increase the dialing rate, then the incidence of such
calls is certain to go up.
Each dialer company will promote its own proprietary method of balancing the
abandoned call rate and idle time for an agent. It is very important to understand
that the abandoned call rate is the ratio of abandoned calls to calls answered
and not calls attempted. It is measured by dialer campaign.
In general, the more calls you are willing to "abandon" the more
"talk time" you can have for your agents.
What is the definition of
an "abandoned call"?
An abandoned call is :
While all dialing products
have a predictive dialing module, the most compelling differences between those
systems and I-Tel may be the manner in which abandoned calls are managed. Our
dialer engine was designed to meet the most stringent codes of dialing practices.
It was designed specifically for a compliant world. With reference to this we
have:
To help you understand how the area of predictive dialer legislation has developed,
the following provides additional reading.
PREDICTIVE DIALERS AND "PACING" METHODS
The productivity of a predictive dialer can be measured by the "predictive benefit"
of the extra agent talk time it provides compared with simpler dialing methods
such as progressive dialing. The quality of a predictive dialer lies in its
ability not just to generate productivity but to do so whilst working under
compliant dialing conditions. In a given period of time (and for a specific
campaign), more contacts will be made when more accounts are dialed and the
wait time between connected calls is kept to a minimum. Abandoned calls and
wait times are simultaneously and successfully managed based on how a dialer
"paces" itself. In the traditional model a dialer will dial more calls
than there are agents, using different methods to determine exactly how many
calls should be launched at any given time. If a dialer over-dials, you will
be forced to abandon calls because there will be agents available to take the
calls. If the system dials too slowly, wait times are increased and you will
be making fewer contacts. Pacing is the heart and soul of dialer technology.
The design and algorithms are technically very complex and the results will
usually separate the effective dialer from all other dialers.
Most dialers have been caught short by the new Federal requirements on dialing.
Whist in most cases it has been, as expected, relatively easy for them to achieve
compliance with the new standards, achieving effective performance under compliance
is something quite different, and a challenge that is proving too great for
most dialer vendors. Most dialer designs originated in a market where there
no rules and consumers were more tolerant of abandoned and other kinds of silent
calls. For example, it was/is not unusual for a dialer to have a guess at the
optimum dialing rate using some mathematical formula, and then go back and forth,
under and overcorrecting as it seeks to find the right balance. Such "seeking"
methods waste abandoned calls in a world where abandoned calls are suddenly
becoming a scarce as opposed to a "free" resource. Federal agencies are clamping
down on abandoned calls and collection agencies may eventually come within this
net. So,
can dialers cope under compliance? Read this interesting document that reviews
the background of compliance and presents several little-known facts about this
subject.
How does our dialer manage
its pacing? We create a "dialing pool" - You could think of this as the area
that we use to launch our predictive calls. When agents become free (not
before), we put more numbers into the dialing pool. The more agents there are,
the more the numbers we add as each agent joins. This is in recognition of the
fact that as agent numbers go up, there is a stronger likelihood that another
agent or other agents will become free to take any overflow calls. Within the
system and taking account of all campaign events such as call
outcomes and talk/wrap distributions, a very sophisticated simulator uses immense
power to calculate the probability of an agent becoming available in the next
second, 2 seconds, 3 seconds and so on. We dial not just for an agent, but for
the way in which all agents are expected to become available. For a given abandoned
call target, our dialer ensures that you get the best possible talk time per
agent hour, or if you like, the lowest agent wait time achievable between calls.
Our design approach is very different from traditional dialer designs.
The results are similarly different.
The
following is an example of a "Campaign status inquiry" for predictive
dialing. Answering machine detection was turned on. Note the small number of
abandoned calls compared to the number of live calls (less than 5%).
The following is the agent status display for a predictive campaign. The color
coding indicates the mode that the specific agent is in (Talk, Wait or Wrap).
For the campaign, talk time per hour is about 43 minutes. Average wait time
is 11 seconds and as the previous chart showed, the abandoned call rate is less
than 5% (33 abandoned compared to 722 completed calls).
In the above, the yellow
marker shows the average length of an event. E.g. an agent spending an average
of 11 seconds in wait mode.
The blue markers show the amount of time spent in a particular state per hour
:
Sytel
has been supplying world-class predictive dialing software to the global call
center market since 1997. Sytel's predictive dialing software was developed
in response to the need to be able to dial effectively at an abandoned call
rate of 1%, or lower. For many years, this need was ignored in the United States,
but is has now become a pressing requirement. Compared to traditional dialers,
I-Tel adopts a significantly different approach with the
Sytel Softdial Plugin®
that uses the Virtual Event Machine (VEM)®, a
highly sophisticated simulator. This is
a software engine running in excess of 10 million simulated calls a second with
blazing speed, constantly adjusting its dialing pace with precision, to match
the rapid changes in campaign conditions.
What is the advantage of this approach? With
traditional dialing solutions, collection supervisors spend a great deal of
time monitoring the dialer and tweaking dialing rates when abandoned calls increase.
A well-designed dialer should allow supervisors to focus on managing their agents
and planning campaign strategies as opposed to second-guessing their dialer.
This common problem does not exist with the I-Tel dialer where you simply
define the maximum abandoned call rate and leave it alone!
For the collection industry we have a set the maximum abandoned call rate at
5% which is higher than the FTC rate but conforms with the DMA's rules. I-Tel
easily copes with all extreme dialing conditions e.g. 80-90% no answers and/or
answering machines. Its performance can not be beaten and we welcome any benchmarking
test that users or other market participants may wish to run. Simply
stated, we give you very high call volumes and low wait times while
meeting your objectives on abandoned calls.