There are two trains of thought within my organisation, when it comes to the caluculation for abandon % and grade of service%.

These are:

To keep the "self abandons" (those abandoned within the 20 second SLA)within our calculations for the above.

Or

To remove these calls as if they never happened, or just remove them from the abandon %

My personal opinion is that the calls should be included in the calculations - thus giving a more accurate picture of performance - if a high number of calls are self abandoned, should we not report these and then seek to see why the callers is hanging up so soon?

I would be interested to hear your opinions on this - is there a industry wide calc for inclusion of self abandons?

Thanks

Ian

IMHO.

All calls that hit a turret and CAN be answered should be counted. Providedthey have actually rung.

Obviously anything that didn't make it to an agent should be discounted.

You MAY also wish to remove anything that abandoned < 2 sec as these also are hard to get.

Again, just my opinion.

HTH

DaveA

I agree with Dave, that if the call gets to a turret then an agent should be able to answer it assuming you are staffed correctly. In terms of reporting, i would say that you should record both figures, if you find that you have an aba rate of 30% and 20% aba within the first 5-10 secs this might indicate an issue to investigate

If you are playing a "Welcome message" before trying to route calls to an agent, your "self abandoned" calls may well be "self serve" calls. E.G. "Thank you for calling the broadband support line, we are currently experiencing an outage in the XXX area and engineers are working on it. Please hold for the next available CSR."

Anyone calling from XXX will have their call satisfied within the first 10 seconds as they have been given the information they require, anyone holding on after that should be counted in the abandons.

Previously I have reported two sets of figures, self serve plus abandons equals 100% of calls not answered.

My preference is to include any that could be answered by the agent. For this to be resonable you would have to look at a couple of things:
- Do you have a ringback time built in? (often 2 seconds between the CDN and the ACD)
- how long does it take from the turret ringing for the agent to start speaking? Factors such as call forcing will come into effect here.

I also like to show both the total abandoned and those that abandoned after the threshold (i.e. after target answer time).

If you find that the aba% is high due to short notice abandoning then these calls should obviously be investigated and highlighted but NOT excluded.

HTH

I think it's perfectly fair to separate the numbers for call abandoned under your target answer time. They need to be reported and investigated.

Can you capture CLI and see if there's a pattern, or even try and contact them if the volume is significant.

- Do they call back later,
- Dial by mistake,
- Think there's going to be a long wait and give up too soon
- Are they customers or is this new business you're losing

Lot's to keep you busy :-)

I agree with John.
Also if you are planning your staffing to a service level of lets say 80 in 20 then you are automatically accepting that you will not answer all your calls within 20 seconds. With that in mind you do not want to penalise yourself for not answering callers who have not given you at least a 20 second chance.

Dave I wonder that how you can get that data ?

"You MAY also wish to remove anything that abandoned < 2 sec as these also are hard to get. "

The calls that abandon within 0-5 secs OK
but specificly how can you get the number of calls abandon before 2 secs ?

Smthing I don't know ? :)

I have worked with dave in the past, you should see him get around the calling floor with a stopwatch, a spectacle to behold if ever i saw one. :P

It's fairly simple,

The CM systems I've used (Harris, MTE, Symposium and Avaya CMS)
all have a report with Time to Abandon normally as a function of
skillset, CDN, ACD, Vector or Application.

This most cases it gives the fields. 'Cals Abandoned'

Then the following time frames.

0-2sec
2-4sec
4-6sec
...

120 sec+

So it's just off a report.

HTH

DaveA

Ok. I've got it

But is it fisible to calculate 0-2, 2-4, etc
I prefer to get it as 0-5, 5-10 and so on

Aydın,

Gathering them in 05,5-10 could be done depending on the system you are using. Symposium I just checked gathers them in 2,4,6,8,10 etc seconds.

Hey,

If the abandone rate has to be minimised the solution is in the IVR,
maybe the IVR announcement is delayed a bit which forces the caller to drop,
or the IVR is not mismatch between customer profile and announcement.
Eg. If we have first announcement which welcomes and if the second announcement which say your call is important to us or any promotional offer is after 30 seconds of hold check if the calls are abandoned at which second? then figure out to place the second announcement at that timeframe.

It's amazing how many companies don't run statistics to confirm when a call abandons.

At one company I worked for the average call would abandon at 25 seconds...yet all our messages would play at 30 seconds...when asked why they picked 30 seconds there was just a collective *shrug*.

So...by changing the message to play at 25 seconds we literally extended the Average Abandon by almost 10 seconds.

This also gives an opportunity to upsell by playing the appropriate IVR!

My feeling is that you need to consider the environment you are in. Our centre is mainly business to business but with a large number of calls from internal staff.

With Sales directors concerned about one of the abandoned calls being "the million pound deal" I did some investigating.

We have an 85% within 30 sec SLA. On average 4-5% of calls are abandoned within the first 30 seconds. However only 0.8% of these are from external customers.

Do internal customers having higher expectations than external customers? The answer is definitely yes in our business.

Hi Andy,

Have you ever heard of a product called Queuebuster.

Here's an idea, with the calls which are over 30 seconds you could say;

Your call is important to us, Informa have invested in Queuebuster technology to ensure every call can be answered.

If you prefer Queuebuster will hold your position in the queue and call you back as soon as it reaches an agent.

This technology is network based and switch independent and available at a reasonable price.

It would certainly set the Sales Directors minds at rest.

I know some people may think this is an add "which it is" but I feel its constructive.

Regards
Jason

As someone who has had experience with a company that uses a "queue buster" type system I agree this is a great method to use.

I don't believe the system I am thinking of was actually branded "queue buster" so there may be a few options out there.

I think the point has been missed here.

Take a step back and remember how many staff you planned to be in situ at a given time and the method you used to calculate how many you needed.

The calculation method will invariably have included a metric based on service level ie 80% of calls answered in 20 seconds. Given this to be true then by default you are only planning to answer a % of your calls within a specific time period, in this instance 20 seconds and you have staffed accordingly.

My belief is that unless you are staffing to answer 100% of your calls in a nano second then those callers who abandoned within your service threshold should be discounted from your service level calculation.
By all means capture the data around callers abandoning within the threshold but use the data to identify areas for investigation.

In summary, if you plan for a threshold of 20 seconds then any caller who abandons within this time period has not realistically given you a chance to answer the call as you have staffed to meet this 20 second requirement only.

When a call is delivered to a peripheral, the caller might be placed in a queue waiting for an agent to become available. Normally, if the caller hangs up before being connected with an agent, the call is considered abandoned. A high number of abandoned calls might mean that I am losing business because callers are being made to wait too long.

However, if a caller hangs up almost immediately after being placed in a queue, I might not want to count that as an abandoned call. In these cases, caller impatience or excessive queue times are not the problem; the caller probably hung up for another reason. Tracking these as abandoned calls can be misleading.

Therefore, I can specify a minimum amount of time that a caller must wait before the call can be considered abandoned. This value is called the abandoned call wait time. I can set this value for each peripheral. A typical value might be 10 seconds. This would mean that if the caller hangs up in the first 10 seconds, the call is not considered abandoned, nor is it counted as a call offered. If the caller waits at least 10 seconds and then hangs up, the call is counted as both offered and abandoned. (In the real-time data, a call is counted as offered as soon as it arrives at the peripheral. Therefore, a short call might appear as a call offered in the real-time data, but is not counted as offered in the historical data.)

So, the formula to measure the abandonment rate % = All calls abandonment / All calls offered (Busy + Received)

And there will be another report to translate Abandon rates to Service level objectives (ATA & Avg. abandon rate when ATA was X)

Example
-------
10 sec. = 3%
20 sec= 1%