Hi all,

I have been given the job of forecasting for a call centre. We are using Genysis, but it isn’t forecasting very accurately. We have a years worth of data and I have high excel skills, so was hoping to utilise this and create a forecasting model to run parallel with Genysis. Apart form the usual TREND, FORECAST etc does anyone recommend “a good way to forecast” or is there a preferred method of excel models?

I’ve looked at the Erlang side of things and obviously it’s very good at the actual running and scenario testing. But doesn’t seam to cover how to build a forecast model from it.

Does anyone have any links, information or just plan guidance to get me started.

As I’m new to this, so be gentle.

Thanks.

Tony,

Before you go and re-invent the wheel, I would suggest understanding why your Genesys tool is not forecasting correctly. The Genesys workforce management tool is fairly mature and I would expect it to be able to cope with your requirements unless they are very complex. I would be looking to your supplier to provide either training or consultancy to sort it out. It would be a shame to be paying large amounts of money on Genesys licenses if you are then going to do all the work yourself manually with Excel.

It could be that with a small amount of analysis and "tweeking" you can get the Genesys WFM working the way it should.

Hope this helps,
Dave

That was my thinking, but the "worker" said it didn't work and that was that. But I will take your advice and go down the offical training route.

Thanks for your help.

A WFM tool is exactly that, the forecasting modules are trend based which are no good if you are in a sales or event driven business.
I think your approach to using excel is the right one as I have yet to see a forecasting tool off the shelf that can cope with everything our beloved CC can throw at it.

I am totally with Steve on this, WFM is for managing your workforce. Most WFM's have a very rudimentary take on Erlang with a few variations added to formulae to give unique results.

IMO use Erlang in Excel to forecast( this will require you to build some models) then overwrite/import the forecast to your WFM and then use its Scheduling function to manage your staff

Well that does change things quite a bit. WFM, no matter what changes are made never gives what I would consider a reliable forecast. I used a basic 6 week moving average in excel and it proved more accurate than WFM. I do take Dave’s point that I need some official training. But the more I’m looking into it, the more I’m going towards creating an excel model.

I guess my next move is to understand Erlang and start from there. I really appreciate the feedback and look forward to any more comments people have.

Thanks once again.

Hi Tony,

Interesting reading your post re Genesys WFM - I would be interested to have a chat to you about you training needs.

It is a good thing that you are running a side-by-side model in Excel - you should validate the numbers any tool gives you until you learn to trust it.

In terms of using Erlang, I would advise caution - especially if you are using some of the more advanced routing features of Genesys to find agents for calls.

Be especially cautious of using Erlang formulae if your agents are in multiple skills groups/Virtual Agent Groups or you are overflowing calls to other teams in the event of high wait times. Also if you are using priorities for calls in queues or Priority Tuning (aka Business Priority Routing) - these features tend to invalidate some of the assumptions the Erlang formula makes - something the Genesys WFM algorithms are built to handle.

Might be a good idea to get some help with Gen WFM - it should be more than capable of forecasting accurately in more complex scenarios than excel can easily handle.

Best regards,

Seb

Judging by the two opposing views regarding forecasting. There seams to be a position of “what ever works” use it. I will create an excel model and run in parallel to WFM and see which gives the more accurate forecast. It would be interesting to know who uses Excel and who uses WFM for forecasting. Maybe the board administrators could do a poll! I think a lot of people would get a lot from it and maybe generate work for training companies etc.

I'd never want you to think the views are so opposing (I actually subscribe to both sides of the argument - talk about sitting on the fence!)

- I would just like you to think about the pros and cons of each approach and understand how valid the tools you choose are for the complexity of your operation... there is no magic bullet here!

Hi Tony,

Forecasting is not as simple as managers expect it to be (especially after the sales talk given by the WFM tool providers .... & Genesys is really good with sales pitch).

Any WFM Forecasting software is not a 'solution' but a 'tool' for foreasting.

I use Genesys for 3 distinct markets and WFM helps me in different ways with each.

I am going to talk you through forecasting for a particular month (hope that is what you were looking for)...

Gen-WFM has an "Expert Average Engine" which works well with minimum 6 weeks of historic I.V. data. This will NOT give you good forecast of daily call volumes but it gives you 2 fundamental factors for sake of your forecasting requirement.
(i) % split of intraday I.V. per 15 minute timestep
(ii) % split of daily I.V. split per week MON through SUN

To get the monthly or 4/6 weekly forecast of total I.V. you will have to accept that it involves 20% of historic data, 15% formulae, 10% excel, 20% Business awareness, 20% Market awareness & 15% of 'gut feeling'.

Do not ignore your gut feeling but make sure you are looking at good (realistic) data before you arrive at a decision.

And keep the calculations simple until you are comfortable with the forecasting.

eg. To forecast calls for a coming month eg. APR-08 you can look at following factors:-

* YoY % change (has it been consistent?)
* MoM % change (this is more reliable in most cases provided other market & business dynamics have not changed drastically)
* use (ii) derived through WFM Exp Avg Eng to split the monthly call figure forecasted by YOU to get the I.V. forecast per day. I have made a decent attempt to do this in excel but i am not sure how i can post the excel sheet here.

Most important, note down your assumptions, calculations for each forecast & compare it with Actual I.V. per day. Genesys WFM itself will help you with this comparison.

You will never be good at Forecasting until you regularly EVALUATE your forecasts against your actuals. Critical evaluation of your forecast will fine tune your gut feeling.

Cheers!

Kex