Fistly happy new year to all.
I'm stuck in the office and have been asked a question about service level.
AS an example I've run some data and have made an interesting conclusion.
We have our service level threshold set to 20 second.
Our WFM app has shown a 0% abandon rate within service level. (Whoopie!!!)
I checked the data as everyone said the app was wrong and low and behold found the Average dalay to answere is 35 seconds. I then had a copuple of strong coffees and re-ran the data for a month. Now we don't report abandoned in IVR as standard. As such the service level and actual Average Abandon Delay are post IVR and Greeting etc so the prediction by the WFM is correct.
What I want to do is adjust the behavoiur of the business so we look at a more realistic Service Level threshold. Where oh where did 80/20 come from? Who is responcible? How old is it and why is it still used?

Are you asking questions or just ranting for a bit? :) If you are looking for testing try just calling the CDN/Queue directly instead of going throug the IVR and see if anything unusual happens. Also, have you checked the average speed of abandon vs answer? It's possible to have a 35 second Average Speed of Answer and still have zero abandons.

What behaviours of the Business are you looking at changing? You want them to start counting abandons in the IVR?

I've personally always thought that was a great idea and am consistently suprised by how few call centres count any IVR stats. I mean...you may think you have an excellent call centre but if you have 20% of your calls abandoning during IVR treatments there is a real problem there. Lot's of call centres never even look though and once an IVR is in place, it's never changed.

Or are you saying there is a problem with a Service Level of 80/20? If so, why is that a problem?

I had a slightly similar experience happen when we tried to determine why our service level was absolutely horrible when we moved to Symposium. Turns out there was a 15 second uniterruptable message which played post-IVR, that means, for all intents and purposes the actual SL was 80/5.

Turns out that from a technical point of view it has (under MTE) been set at 80/35 but everyone had forgotten that many years ago.

Okay, after re-reading your post a couple of times I realized you were saying that an 80/20 SL is the behaviour you want to change.

What you didn't include is why. What's not realistic about this goal?

At my particular company many of the different departments have an 80/20, in fact generally the more "important" one's do. I don't see any inherent problem with trying to reach that goal.

Sorry if I didn't make it clear what I'm after but the first day back in the office and my brain wasn't functioning correctly.

What I want to know is the history around 80/20. WHat is is based against, where did it come from and what data was used to produce it.

Many thanks

Nick

What I want to know is the history around 80/20. WHat is is based against, where did it come from and what data was used to produce it.

Ever Call centre is different, at mine we figured that answering 80% of our calls within 20 seconds was an acheivable goal and would increase customer satisfaction, so we did it.

80/20 I wouldn't say is "standard", not sure if there is a "standard" servicel level. I've always encouraged every department/company to decide what they want, what they can do and what their customers will tolerate.

I would echo Justin's comments, the SLA you set is based on what your competition work to, what your customers will tolerate and what you can afford - the more challenging the service level eg 90/10 the more idle time you need to achieve it and therefore the more staff you need to be available.

Maybe the 80/20 figure itself was borne out of the Pareto principle that states that for many phenomena, 80% of the consequences stem from 20% of the causes. In business this idea is often applied to data such as sales figures: "20% of clients are responsible for 80% of sales volume".

Therefore in early callcentres business consultants probably thought in the absence of any other data this approximation would equally apply to service levels and what would satisfy most customers.

is pretty middle of the road and should produce around a 3 or 4% abandon rate. I think that the "correct" service level for you is one that:
*Meets customer needs and expectations
*Keeps abandonmement at an acceptable level
*Minimises agent burnout and errors
*Minimises cost
*Maximises revenue
*Is agreed upon and supported by senior management

I have never understood why this always remains fixed. Centres work year on year to a Service Level that most either achieve quite easliy or never get anywhere near. Surely the goalposts should always be moving.

If you have achieved 80% within 20 seconds last year why agree not a new SLA with your business for this year! Even if it is 81% within 20 seconds. Shouldn't we all be looking for continual improvement?

If you are over achieving your Service Level then yes you could up the target. But going back to the other reasons for setting a standard in the first place you would only do so if the conditions suggest that a rise in SLA target will be beneficial to organisational objectives.

In truth exceeding your target service level means you had too many staff and if your objective is to make a profit then you will cut staff numbers to maximise profit rather than just raise the target to return to equilibrium.

Does it follow that exceeding the service level means a cut in staff? Could that not be a prompt to win more business? Especially since the new business win be so much easier to win given the excellent and improving service levels.

Also there is (perhaps) an implicit assumption in this discussion that the only obvious way to improve service levels is to have more staff. How about improving the system of work so that people stop ringing in to query, chase progress, complain or similar? How about if people only called to buy, upgrade, change address and for other valuable reasons? I bet that in any call centre in the world over 50% of calls are to deal with failure. Stop these calls and you get your idle time and service level for free at the same time as you improve the up front service for your customers.

Maybe staff know that is they exceed the SLA there will be cuts in staffing levels. Maybe that is one of the things holding service levels back. If they knew that exceeding the SLA would give them more and more varied work, they then might muck in to help.

Per my response, no over achieving does not necessarily lead to a cut in staff, constantly raising the bar on speed of answer may be the goal of the organisation. However, if your current SLA is determined to be the benchmark within your particular area of business then why raise it unless there is some particular business gain over your competition which outweighs the lost profit of not reducing FTE.

Simple contact centre mathematics determines the optimum distribution of idle time and resource levels against a targeted service Level based on your expected AHT.

If you are achieving a better SLA than your target then your agents are idle longer than they need to be which means you have inefficiency. If this can't be filled with other new work then you need to cut staff in your callcentre.

Maybe staff know that is they exceed the SLA there will be cuts in staffing levels. Maybe that is one of the things holding service levels back. If they knew that exceeding the SLA would give them more and more varied work, they then might muck in to help.


I just don't think they conciously do this.

There are many examples where employees will not exceed a quota or standard because it would be detrimental. My grandfather used to work for a ceramics company making baths and toilets and the like. He was told in no uncertain terms by the the other staff to slow down as he was making the others look lazy.

And Julian says:

exceeding your target service level means you had too many staff and if your objective is to make a profit then you will cut staff numbers to maximise profit

If we know that do you think that the staff don't?

I will concede that it won't be like that in every place, but it will in many. Hence the "maybe" get out clause.

I worked for BT in sales about 8 years ago now. One day they dropped the comission values so far that I desided to have a change so I went to work in the complaints department.

The complaints department I worked in only responded to complaints using call recording and were not customer facing.

During my first week I set about clearing the huge backlog of complaints left by the pervious person. Being a salesman I set myself a target and to the dismay of my colleagues, set to work. Within a few days one of the self appointed senior members of the team took me to one side and explained that I was making them all look bad. He also said that if we didnt have a back log there would be no point in working overtime on Sundays which was either time and half or double pay.


I learnt to play Howard Marks online game and sat it out in that complaints
department for 6 months before I finaly got bored and left.

Thanks to ICMI for this one..

While I have no research to prove it, my theory on the popularity of the 80/20 service level stems from our handling of incoming calls way back when calls were delivered directly to agents rather than to an informational recording. In lieu of messages informing callers of our monitoring program and how much better it would be if they went to the Web for help, we had no choice but to deliver the call to a human being (or to a queuing message, if all agents were busy). To "buy a little time" when agents were not available, we would send the caller four cycles of artificial ringback. Each ring cycle lasted around five seconds, so this strategy would buy us 20 seconds or so to free an agent. In other words, from the caller's perspective, any call answered in 20 seconds or less sounded like an immediate answer. Therefore, our 80/20 goal translated into four out of five callers thinking that the call was answered immediately.

It's been a while since this topic was live but I wanted to share this short thought from William (Bill) Durr with you.

Higher is not better, it is not like profit. There is an acceptable range for the metric, much like tire pressure. Too high or too low is dangerous. Service level is a negotiation of sorts between speed of answer and cost. Don’t make the mistake of seeing service level in the same light as profit. Generally, if you over-achieve on service level you have actually incurred higher costs than you were required to incur.

Hi,

I'm new to this forum and to the ins and outs of call centre management but have noticed something about 80/20 service levels...

The 80/20 SL appears to drive the wrong management bahaviour. This is because when resource is tight, the preference is to flood the phones with people when volumes are highest achieveing sub-20 seconds wait times on the basis that we can live with >20 second wait times the rest of the time - and still meet the 80/20 SL. This seems very non-customer focused?

I believe that customers feel this stark contrast in service - especially when the multiple inbound calls are being made by the same customer. Should we not focus on achieving a 'stable' service, where calls answered too quickly or too slowly are considered 'defective'? This would drive a more balanced approach to staffing and deliver a service that conatins much less variation?

And if this is the case, then the toolset for delivering this is one that employs 'control' which actually relys no 'target' at all but instead focuses on reacting to real-time variation in queue volumes etc?

I would appreciate any advice on the subject.

80/20 or whatever SL is intended is a pre-requisite for the erlang calculation, ASA after the event is the sum total of your operational performance.

Technically 80/20 can be achieved with an average speed of answer in excess of 20 seconds, more likely though the ASA will be nearer to 7 or 8 seconds.


As for it's origins I don't know, but on the face of it it seems like a reasonable target and one that is affordable and one that customers would accept (If they knew).

It is interesting that this subject has run for nearly a year now. Early in '07 we reset our service level targets in line with the correlation to abandon rate and hence none of our "service" lines have an 80/20 target.

Although we have for some time set a target service level which varies by inbound line, depending on the customer's propensity to abandon, I am under increasing pressure to deliver to target service level by interval. This will help to delvier the "stable service" which Nathan suggests. This is clearly a customer centric approach and means that we will attempt to answer every call within the known tolerance of our customers.

However, I would appreciate views on the practicallity of attempting to deliver service level by interval without incurring increased costs due to over staffing. Has anyone ever achieved target service level every interval of the day and what conditions are required to achieve it?

Mike