Morning all,

Just to open up a little debate.

Ignoring Intraday for the moment.

At what point do you re-forecast your long term
model based on apparent variance? Do you have a trigger
level set that automatiacally requires a re-forecast or
do you work dynamically?

Personally I like (and use) a dynamic model so if there is a
consistant x% drift I'll re-forecast rather than waiting for the y%
trigger, although I'm open to opinions.

Regards

DaveA

I would check my long Term Forecast every single day to analyse upword/downwords drift. I will be worried if the variance beyond is +/- 5%. Since LTF is more or less for budgetting and hiring requisitions I would consider those (Procurement, Hiring, Training) lead times to revise my forecast or atleast raise a red flag to all the concerning departments to make the amendments.

Hi Guys,

Care giving your best practices on LTF? If you dont mid :-)

I can establish seasonality, but how do i project annual volume? Say i have 2-3 years worth of data.

Thanks,
Mike

My method of developing LTF is based on month over month % change in volumes for last 2-3 years. It not only outlines the seasonality but also shows year on year product growth. In a way I dont forecast volumes but % change and apply to last actual volumes. I use weighted average to derive the % to be applied. I also take in consideration factors such as new/lost business volumes and any changes in factors that may directly/indiectly impact product forecast.

I see. So your next year's projected volume would be the cummulative of the month over month % trend you've established for the 2-3 years data worth.

Thanks for the help!

Dave,

It depends for me.

1) What has brought about the variance?, is it trending due to a particular activity?, is it seasonal?.

My answer would be it depends on the particular situation.

Do you have an example?

Tim

Dave,

if the forecast is based on the regression modelling or has the correlation developed then you would also need to keep an eye on the change in correlation or the change in the other independent factor, which may impact your LTF.

For example:-

In telecom sector the Inbound Volume at CS is dependent on Call Per Subscriber. If due to the change in advertising campaign or any other reason the subcriber base increases at much higher levels, then it will impact your LTF for a month/Year.

Hope this helps.

Thanks,
M

Hi Dave,
Not including Intradays I check the forecast on a weekly, monthly and quarterly basis. (and then again when it comes to the annual budgets)
I only tend to make changes if the variance is 5% or more, and not down to a one off event.
Events being mailings or press releases etc that would affect the calls that day or over the course of a few days etc.
The weekly check is more to make small adjustments as we see how a month developes.
The monthly and quarterly checks enable me to identify developing trends and make adjustments to future budgets where required.

I look at my LTF once a week as it forms the basis of my daily and intraday forecasts. (dauly and intraday are looked at several times a day)

If you are using a regression model you should be able to establish the weighting any event has on the final number and pre-emt exceptions to the normal detrended forecast. Any variance after that can then create a warning light to either evaluate changes in your variables or the change seen in the detrened volumes.

ie marketting email sent should add to predicted volumes. Previous campaigns can be analysed for impact above detrended volume. This will allow you to gage what increase is normal and when it becomes exceptional.

LTF as we understand means Long Term Forecast. While I undestand the need of intra-day and daily checks on variances to tweek the forecast it will not serve the purpose for the LTF. LTF entirely is required for Budgeting and hiring perspective. You wouldn't want to change the hiring numbers every week on slightest of the variances.

It about management dicretion to keep some fat while hiring as proposed by LTF and then skim it during scheduling and shrinkage management.