I'm currently running a soft dialler and all conversations always steer towards AMD. I've tested it on the current system and it's appallingly bad and as such it's a major driver to upgrading. All hard dialler manufacturers quote 98%+ reduction in answer machines to agents, however this in my understanding is simply a measurement of how many calls were no longer sent to agents and does not include any measurements of how many of these 'detected' answer machines were in fact real customers that were misclassified??

Does anyone have any experience in measuring this?
Any ideas would be welcome

Hi Nathan, and welcome to CCV.

The usefulness of AMD detection is extremely limited nowadays. That's not the story you'll get from most dialler suppliers (because they think it's a good selling point), but it's the truth.

The main reason is that AM's are now mostly network based (think your mobile voicemail or BT answer) or they are digital home systems. Neither of these have the telltale clicks and pauses that you got with tape based answerphones.

Failure to detect may annoy your agents, but a false positive is even worse. How many silent calls are being caused by AMD?

I've just been looking at the new Mag. North dialler which isn't running AMD for, I understand, exactly these reasons. Their alternative is to have an "Answerphone" button for the agent to click as soon as they realise they've got one.

It might cost the agent a minute or two talk time in terms of pressing the button, but it'll actually mean you can dial a bit harder because you won't be making silent calls because of AMD.

Of course, that's assuming that you were correclty reporting AMD false positives as silent calls in the first place.

Regards,
Darryl

Thanks Darryl - you've echoed my sentiments exactly, it's interesting to finally hear a similar opinion from another party - thanks a lot for this. I've had various conversations with some of the bigger players in the dialler market and they all seem to avoid answering this question - I'd be interested to see how they try and sell it if the OFCOM proposals take full effect next year!!

I'm keen to gather views from all over so if there is anyone who has a different view to Darryl or myself I'm sure we'd both be interested to hear.... anyone??

Lets look at the benefits of automatically screening answering machines from an agent in a call centre that places 20,000 calls per day.

Assuming the following:

Call connections = 30% or 6,000
Answering machines = 25% or 5000
Busy no answer = 45% or 9000

Let's assume that the handling time for each answering machine passed on to the agent is just 30 seconds. These 5000 answering machines a day represent 2500 minutes or 41.7 hours of lost agent time per-day at 312 calling days per year that's 13 hours of agent time annually at $10 per agent per agent hour this represents $130,000 per year spent on answering machine handling.

I don't have an answer to your original question regarding AMD and registering connection calls as answer phones. I haven't been asked the question before so cannot comment but I don't believe it is a problem at the higher end of the dialer market.

I don't agree that the answer is to remove AMD all together.

Jason

Jason,

I have no doubts about the cost benefits of screening answer machines automatically, however from both a customer services and best practive point of view I want to be able to understand and measure the amount of answer machines misclassified, potentially by NOT being able to report accurately on your abandoned calls are you by ommission breaching DMA/OFCOM guidelines??

I refer you to my original question I want to see if anyone else out there has successfully managed to report on this figure, the only way I can see it being done is to record and playback a selection of dialler calls (not delivered to agents) - however the selection, in my opinion, should be fairly large, spread over a number of days dialling at different times onto different networks - I don't fancy running this kind of test if someone else already has!

...but they're assuming that AMD works properly.

I don't believe it does - not in any PDS.

The impact of it not working properly is that call centre's make silent calls which aren't accounted for. This means that many CC's, who believe they are running at 5%, are actually running significantly higher.

The numbers are totally misleading and it is surprising that no one has pointed this out. For example agents can easily dispose of answering machines in a few seconds.

There are really 3 issues here.

1. The Rules. The current UK DMA rules say, and have said for some time that if there is agent to connect to then a call should be hung up after a second. What this actually means is that UK call centres should not be deploying answering machine detection, if they wish to abide by the rules. Ofcom have yet to address this issue but it will be surprising if they set anything that different.

2. Common Sense. In the UK many users turn off AMD for the very good reason that when they do so their sales per hour go up; yes up and not down. The reasoning is simple. When AMD is turned on, talk time per hour goes down by a few minutes, but is more than compensated for by improved quality of call. Think about it - if you keep someone waiting while you figure out whether or not he/she is a machine before you can be bothered to talk to them, then how do you expect them to react? Some users are so oversold on AMD that nothing will persuade them of this. Well, the approach then is to invite them to do a trial; turn it on for a week, and then turn it off in a controlled trial. That is our advice to users where AMD is not against the rules, and turning off always wins.

3. Dropped live calls. As at least one reader rightly notes AMD will drop calls answered by persons, mistaking them for machines. The quicker you try to do it, the more likely this is.

In the UK Ofcom are aware of all these issues and it let's hope they give a good lead in their final rulings so that consumers get a better deal.

I have to agree that it does not take 30 seconds to dispotion answering machines.

What I have done before on different dialers is make the AMD less aggressive. Often, this would allow those calls that would not pass through to finally get connected to agents. Sure, doing this also let more answering machines through but by making adjustments you might fine the proper ratio that works for your dialer, list, time of day or geographic region.

A test I also like to do is run my staff with answering machine detection turned off. By doing this, you can see the ACTUAL numbers you should be getting.

I have looked at this discussion, my two bits to it !

We use a dialer that has given us the facility to determine the AMD detection time and resulting accuracy. This offers us a more realistic option of correctly tuning the detection to a level wheree detection is fairly accurate but does not take so much time thatw e lose live calls.

Rather that totally discard AMD this is a better option to allow the user the facility to determine the AMD rate, and resulting better control over the ops. A tradeoff works better.

Anyone see this facility on other dialers?

Dadi

The point being missed here is hard vs soft diallers.
With a soft dialler the AMD is not being performed by the dialler but is a function of the switch. As a result it is highly unlikely this is ever going to be as effective, or tuned, as the AMD in a hard dialler.

It's true that a soft dialler can only be as good as the exposure it's granted to the underlying hardware (which may be minimal) - but to suggest that a hard dialler is still any good at AMD is poppycock.

No AMD is effective enough (see my original post) to warrant being used (see my subsequent post)- if you think your product really is good enough then please let me know. I'll be more than happy to carry out an indepedant evaluation of it and I'll publish the results in a well know UK call centre publication. I'll also happily eat humble pie if I'm proved wrong.

But until then, I'll finish by saying that what's even more annoying than crap AMD is crap suppliers saying that their crap AMD is good.

Regards,
Darryl

P.S. Welcome to CCV Joanne! Sorry my reply to one of your first posts is so forthright ;-)

Hi Darryl
I work for an independant systems integrator who has implemented and currently supports a number of diallers, both hard and soft, from a number of different vendors.
We have major corporates using AMD very successfully (on hard diallers) and receiving considerable financial savings as a result, and your suggestion to not use such functionality I find absurd.
I would suggest it is more about understanding your business, your customers, the required tolerance levels and then tuning the dialler appropriately.

Question Joanne do you represent Datapoint the Avaya Gold partner!

If not apologies if so how does being an Avaya reseller make you an independant systems integrator.

Loaded I know and I will probably wish I hadnt asked that one.


Jason

Hi Jason,
Yes I work for Datapoint. We are an Avaya partner, but we are also partners with Concerto/Aspect, the manufacturers of the Conversations, Unison and EnsemblePro diallers, and have also worked in the past with MagNorth.
We are therefore in a position whereby we can evaluate the customers requirements, and assess their total technology strategy before recommending a solution, and are not simply being driven to make a certain product fit.

You can always trust Jason to spot an Avaya reseller!

Joanna, where to start?!....

So you're telling me that the products you sell have AMD that works well?

It doesn't have any trouble with digital or network based answerphones?

It results in no false positives?

You say:
receiving considerable financial savings as a result

I don't doubt that. Last time someone told me that it turned out that the AMD had been cranked up to ensure that no answerphones were connected to agents. On paper this looked great because productivity increased.

In reality, because AMD is unreliable, there were a lot of false positives. This means that real people answered the phone but the calls were dropped because the dialler thought they were an answerphone. These calls were not counted in the dialler reports as "silent calls" - but in fact they are. You get a lot of annoyed customers, and you'll get into a lot of trouble with OFCOM for making too many silent calls.

Any salesman/"independant systems integrator" who says that AMD is effective reminds of the guy a couple of years back who told me that their dialler was getting 200% increase in productivity. Later on in the meeting I found out that the case in point was running about 15% silent calls.

We have major corporates using AMD very successfully (on hard diallers)

I assume that means you're accepting my challenge? I look forward to eating humble pie.

Hi All,

This is my first post so i apologise now if im commenting on old ground.

I work with a really old centenium dialler and cant make jack changes to it ok i dont have a clue about how to go about it. I also dont yet have the tech skill to make and measure any major changes i want to do. "i kinda got flung into this" But looking back on Jasons earlier costing quote I agree. Apart from my answer machine rate was higher. I just reduced the number of rings inkeeping with Ofcom and then measured the difference in actual connects and hit a bench mark where even though i lost connects, ive cut out more AM's so i get a higher connect rate anyway. Does this make any sense to you guys? or am i back in the dark ages? I assume that the systems you use are a lot more effective than mine. I think the dialler at my site had a upgrade about 6 years ago! But maybe going back to basics might increase the general productivity of the dialling campaign. Or do you think thats rubbish?

i agree with all those in this post that say amd is important
i will give a very simple example of this

we purchased data for 20 thousand records contacting mobile phone customers .
with our prev soft dialer that had no amd this is the results we got

answer machines = 50%
live connects = 30%
dead numbers = 20%

we ran the same data through our new hard dialer. that has amd

answer machines =10%
live connects =70%
dead numbers =20%


please note these results are based on an hourly basis per agent.
it works better when contacting mobile phones than when contacting land lines . the reason being the audio signatures are easier to detect on mobile phones than land lines .
when contacting mobile phones it is only natural people have their phones switched off or on silent during normal working hours and these calls always go to answer machines, it is therefore common sense to have amd in your dialer if you are contacting mobile phones..
i,ll leave on a simple note . the people who developed this amd technology did so for a reason??????????????????????????

please note our dialer meets all ofcom silent call regulations ,

hi everyone,
could anyone please help me understand what is meant by trunk line overflow?
we are using a power dialer because we only have 5 agents running a test campaign before increasing the number.the dialer is a hosted solution.

Paul Young

A trunk line on a dialler is usually a group channels connecting to the PSTN or tie trunks connecting to another switch.

An overflow usually refers to calls your system has sent down an alternative path, i.e. in an inbound scenario if no agents are available on original route or all channels are busy.

Your system provider should be able to give you an accurate description though of what it exactly means on your system.

thanks John.much appreciation. could you please help me so that i can be able to reduce the trunk line calls?

P