A password will be e-mailed to you.

As we know, promotional forecasting is a challenging area for all Demand Planning professionals.

Few factors often responsible for that challenge are:

  1. Irregular demand – as many products are very seasonal and simply not available for sale through the whole year,
  2. Unprecedented discounts/ price levels – in a promotional business, we often want to surprise our customers and position our offers as the “best they ever saw”
  3. Much bigger contribution of new product sales to total revenue vs. traditional retail models
  4. A very short life cycle & at the same time, very high number of new products introductions each year
  5. On top of that – our clients are getting used to an idea that they can cherry- pick among different promotions & as a result, we observe very high cross-cannibalization between these promotions even in totally different products categories

As a result of the above, a straight forward approach to a traditional statistical forecasting at a product level doesn’t bring the result we would expect, even though, most of the IT companies selling Demand Planning software, are claiming that their set of algorithms is able to crack the problem.
[bar group=”content”]

Have you been in that situation before, still frustrated with results and looking for a new or different approach?  Yet, at the same time, a practical and cost–effective approach on how to improve your forecast accuracy levels?

Well, during IBF’s Supply Chain Forecasting & Planning Conference in Amsterdam, November 18-20, 2015, I will be discussing 2 algorithms which may solve your problems, while identifying up to 30% of your forecast outliers.  Of course, if used properly, they can ultimately help to reduce your inventory levels and improve profitability too.

During the IBF Session you will learn:

  • A totally new statistical approach to predict and manage forecasting outliers, using logistic regression
  • A new automated approach to prioritize products and planners time, based on the nature of an error (an alternative idea vs. FAV quadrants)

These algorithms could later be modelled using MS Excel & one of the free-ware statistical packages widely available on the internet. So you will not need to buy any expensive software to test this approach at your company.

As a bonus, we will discuss, what skill are exactly required to run these types of modelling and how & where to find them in a market.

I look forward to meeting you at the IBF Conference!