Predictive dialers produce higher agent talk time per hour by dialing multiple lines (‘overdialing’) each time an agent is available.
Regulations were introduced in the UK and US in order to limit nuisance calls to customers caused by this overdialing. believes that the design of these rules does exactly this. It’s pretty obvious that if you inflict significant levels of nuisance calls on your customers, no matter what the market vertical, then you are turning them into enemies, rather than building them up as friends.
And choosing to adopt a fully compliant policy, and being seen to do so, then that is just great news for how an organisation is perceived in the market.
But dialer designs vary widely. Some are very basic, e.g. where the overdial function is just a ratchet on the supervisor screen that can be raised or lowered by the dialer supervisor as they deem fit.
And at the other end of the scale is a dialer engine such as the Virtual Event Machine (VEM®), provided by the , which seeks, using highly automated AI procedures, to produce the best possible dialing result under compliance with FTC/ FCC rules governing the use of predictive dialers.
What are the US rules for predictive dialers?
There are three basic rules in the US. You may overdial, provided that you comply as follows:
Rule type | Detail |
|---|---|
| Early hangups | Each called number must ring for at least 15 seconds, before any hang-up by the caller. |
| Abandoned calls | Abandoned calls measured in respect of ‘live calls’ and not ‘all calls’ should not exceed 3% of live calls, and where the live call measure is… …aban calls/(aban calls plus live calls). |
| Live customers held in queue | Once a call is connected, the phone goes off hook and the customer starts speaking, then the call should be held in queue for no more than two seconds after the end of the customer greeting (this is FTC rule; the UK rule measures the two seconds from start of greeting). |
Let’s take a closer look at each of these elements…
Early hangups
Going back to the 90s and the early part of this century it could take up to ten seconds to reach a called party, because of latency and delay issues, and during at least some of this time you could cancel a call before it started ringing, thus not causing a nuisance. And you would do this if the agent(s) suddenly got another live call(s). But life has changed! You pulse the digits out to the network and with today’s dialers and networks you are ringing in the called party’s home in round about a second a two. So, if you abort a call early, before the called party can reach the phone, it becomes a nuisance call.
The scope for any productivity gain, in doing so, is extremely small, and the chances are that you are actually hanging up on a number that is already ringing. All regulators that we aware of, have ruled this practice out with virtually no discussion, on the grounds that it can be extremely upsetting to customers.
Some years ago, when the US Direct Marketing Association published its own code of practice, it decided that a 12 second cutoff was reasonable. This was later changed to 15 seconds by the FTC when it ruled in 2002. In today’s markets, for calls to cell phones, there is a case for industry to suggest going back to a lower number, but we are not aware of any pressure to do so.
Our recommendation is that the 15 second rule should be observed.
Abandoned calls
Many global US banks calculate this measure as follows – (abandoned calls plus aborts)/completed dials. This is the ‘all calls’ measure and has been rejected by all regulators. Some vendors still encourage their users to adopt it on the basis that it sustains a much higher overdial rate, but it does so only at the expense, particularly these days, of a lot more abandoned calls. Take for example a campaign with a live call rate of say 20%, then the ‘all calls’ measure of say 3% translates into an abandoned call rate of 15%, when the ‘live call’ measure is used.
(As an approximate measure to make the conversion to the ‘all calls’ rate, simply divide 100 by the live call % and then multiply the ‘all calls’ rate by this number.)
This issue has been the subject of much discussion over the years, and to the extent that global US banks take their lead from their vendor(s) as to what measure is acceptable, it is surprising that the ‘all calls’ measure is being encouraged.
Our recommendation is that the ‘live calls’ measure must be used.
Live customers held in queue
Non-compliance with this rule is almost certainly the biggest contributor to nuisance calls in all territories, where predictive dialing is used. The point of the rule, as far as the regulators are/were concerned, is to minimise the nuisance by then playing a message such as “Oops we slipped up. We called you but don’t have an agent available. We will call you back shortly and make sure that an agent will indeed be there to talk to you”, or similar, and then hanging up, i.e. abandoning the call.
Instead of doing this, we see contact centers either allowing dead air, or playing a message asking customers to hang on, since an agent will be available shortly.
It perhaps takes a special sort of person who is happy to be treated in this way.
We have always thought this is not just a bad practice but one that is not in the interests of contact centers. We conducted our own experiment some years ago and made a 100 live calls on a random basis across the US, just to see how customers reacted. We asked our caller to say nothing and just record the point at which a customer hung up on what they saw as a dead air call. And we asked our caller to take a note of anything these customers said – to the dead air line.
Conventional wisdom is that people hang up early. Many of us say that is how we behave. We want to show that we have no truck with silent calls and kill them quickly. Ask anyone over a drink down at the bar how they behave, and this is it. We all want to show that we won’t be messed with!
But it simply isn’t true. When you call somebody, people genuinely want to know what is going on. Could be really good news. Or maybe you haven’t talked to someone today, so you hang on. In our test, the average hang-up time was 13 seconds! Now dialer vendors know this and this is one reason why their own customers are encouraged to do it.
But the real issue is what happens while a customer is listening to dead air? Happy to wait? Not at all. And playing a message instead of dead air may make little difference. The live callers we contacted were very unhappy and subjected our lady doing the test to some serious invective.
So the point of all this, is that to ring a customer and keep them waiting is a straight intrusion on their privacy. Customers resent it and it is certain to interfere with the quality of the call.
Our recommendation is that the FTC call abandonment provision of two seconds after the end of the greeting should be followed.
Can dialers manage to dial productively under compliance in todays markets?
In most markets today, outbound campaigns typically encounter much lower live call rates, than was the case, say ten years ago. And, paradoxically, this means the potential scope for predictive dialers is much greater than ever. In other words, the predictive gain goes up – substantially.
And the acid test, the only test, of any dialer’s performance is its predictive gain under compliance. This is the difference in talk time per hour as you move from progressive to predictive dialing.
Whilst compliance makes sense in terms of not causing nuisance calls, there has been little serious analysis as to whether dialers can dial effectively under compliance. It is certainly true of both the UK and the US that the dialer community was essentially a ‘no-show’ when it came to deciding with regulators, for example, what a reasonable level for abandoned calls should be. And the result is that in the absence of any substantive dialogue between regulators and industry, regulators have erred on the side of caution i.e. they set a very tough level for abandoned calls. And industry by and large hasn’t complained. Rather, many vendors openly support compliance rules and their adherence to them.
Forgive our frustration, but this is an Alice in Wonderland World, where reality and non-reality are sometimes indistinguishable. The brutal fact is that , almost certainly, still has the only dialer that was built to deliver very good performance under compliance. And this matters, because if a dialer wasn’t built to do this, then the only path to very good performance is to dial outside compliance.
Our strong message is that US organisations can expect to dial under compliance with the and get good results. And this means no nuisance calls except just a few abandoned calls. You don’t need to abort calls early and you don’t need to keep customers waiting in a hold queue. As we noted earlier, it’s not just that these actions are outside compliance rules, they simply upset customers.
Are there compliance exceptions?
Yes, but they should be few and they usually apply to the collection of very old debt. Typically debt lists sold and resold in the collections market where the live call rate is likely to be between one and five per cent only.
But in the case of most US collections operations, we would expect that most campaigns would have a minimum live call rate of 10% and a reasonable number of agents, say 20 upwards. At an abandoned target of just 3% of live calls, the AI Dialer will give good predictive gain, and it will improve as these numbers rise.
Nevertheless, accepts that in territories where there is no notion of compliance, and live call rates are very poor, there may be a case for raising the abandoned call target beyond 3%. We think these instances are few. We always work with users in such cases to start at 3% and, in the unlikely event that the results are poor, move to a higher target, normally 10%, again measured as a % of live calls.
