Unique situation: Our client requires 11 languages to be covered during his operational hours. Each agent can speak 2,3,4,5 or even 6 of these languages in any combination.

How do I calculate the number of resources required for this client?

Tx

Eric,

Do you have a WFM product?

If so I'd suggest setting the language skills as if
it were a 'traditional' multi skilled environment, then, just
let the WFM application do the hard work for you.

Obviously this is dependant on having:

a) The historical data available, BY LANGUAGE, to
properly configure the model

and

b) Enough staff to cover the outcome!

Can't think of an easier way to do it I'm afraid, manually is
going to be nigh on impossible.

Another thing to bear in mind, are you working internationally?
If so you may need to offset language skills against the
time zone(s) of the predominant demand, ie: Arabic/ Middle Eastern
'core' business hours are approx 5hrs ahead of the UK, so that's
3am- 1pm in the UK. Turkish would be 2 ahead etc...

It's an interesting question, can you let us know how you get on?

Regards

DaveA

Hi Dave,

WFM Product - None. I have designed a Workforce tool in Excel which is extrememly effective ( even one of my managers who worked on Aspect for years state that it is easier to manipulate schedules than aspect )

I am trying to get access to a WFM System, but because the servers etc sits in India, access is not that easy to come by.

I have enough historical data, I have it per language per interval and have calculated requirements per language. That was the easy part. It is the combination of all the different languages that agents have, which creates the chaos.

Simple way would be to calculate resource requirements per language and add them all up and that is it.... but you might sit with over staffing

Simple way would be to calculate resource requirements per language and
add them all up and that is it.... but you might sit with over staffing


Exactly, and by a BIG margin!

You'll lose the economy of scale regarding the multi language
skilled agents in that case.

Unfortunatley, whilst I'm a strong proponent of manual
planning tools, and, non over reliance on WFM models I personally
think in this case gaining access to the WFM application is really
going to be the way to go. Eleven parallel skills with upto six
skills per agent is really going to be over ambitious for multi queue
manual modelling!

Sorry I can't offer any easy way out here.

Regards

DaveA

Tx Dave.. I agree that accessing a WFM system will be the way to go.

But I guess I have to do the best I can with what I currently have.... it has been working so far - over 1500 agents on a manual system... easy :)

Hi Eric

Firstly congratulations on creating an Excel WFM tool that copes with 1,500+ agents and 11 languages, that's no mean feat!

Dave is right, a WFM tool would simplify a lot of this but as you don't have access to one at the moment you'll need to make some estimates around how efficiently your operation works. This is how I would approach it.

Working out the resource requirements by language at interval level is a good start point. Adding all of the individual language requirements together would show you the Maximum resource you would need to support each language individually. This is more agents than you need but does provide you with the basis upon which to work out the actual agent requirement overall and by individual language.

Look at a past week where service levels were achieved (but not greatly exceeded). Use the actual contacts and AHT's to work out how many agents would have been required were each language to be serviced individually and compare that to the number who were actually employed in that week. The number actually employed should be proportionatly smaller than that worked out in the individual language calculation. This will give you an efficiency factor = (total agents required by language/total agents employed) - 1

lets say this comes to 10%
e.g. (1,650/1500)-1 = 10% efficency

You can then assume that the total forecast volumes, worked out individually, will require 10% less resource overall.

For individual languages it is tougher. You could use the efficiency factor for each individual language across the board. If the number of contacts per language is reasonably even then this would work well, I'm assuming that it isn't and there are small language groups that need to be looked after in terms of meeting their service level, in which case you'd need to assume a lower efficiency level in those groups balanced by a higer efficiency level in the larger groups.

It's hard to give you a greater steer without seeing the volumes and skillset of the agent groups, hopefully this gives you a starting point. Happy to help further if you want to follow up on this.

Eamon

This is a general question. I agree with both Dave and Eamon. But if you use a Fractional agent calculation in erlang and add all thos together will you not obtain a real total requirement?

But if you use a Fractional agent calculation in erlang and add all thos together will you not obtain a real total requirement?

Not exactly,

Fractional calculations will work out what is needed for the partial
queues, however, if you are not looking at the global structure of
the delivery skills you'll end up with several 'micro' queues that
will take a set percentage of your FTE availabilty, this leads to the loss
of economy of scale I described above.

It will not give as high an FTE requirement as mapping the languages
individually, but, may lose some of the cross over cover.

Regards

DaveA

What are the oprational hours for the business ?

No doubt,tackling multi skill is a tough task but I believe the start point has to be individual skill requirement based on respective volume patterns and AHT. After that you may want to combine same skill workload based on combined arrival and calculate HC req based on weighted AHT. The key here is rostering and then doing Real Time Management to ensure atleast 1 FTE is available per skill at all times. To ensure that, you may also assign Skill levels to the agents and give them priorities for ACD to determine. No doubt your process would be overstaffed but that is cost you pay to service such multiskill process and hope you bake that cost in your revenue.

Looks like too much theory but you will achieve best results once you start implementing and do some tweaking.