Are forecasting errors causing
costly stock-outs and excessive inventory?
Because of several non-IT issues, even the best
forecasting/planning software cannot improve accuracy without addressing
issues such as:
the “bullwhip effect” which amplifies forecast errors upstream along the
other than demand that influence sales, e.g. stocks, prices, promos, batch
ordering, target deadline spikes etc.
data load of forecasting & planning (large systems become unwieldy;
forecasting/planning systems must be lean)
The Dubai Supply Chain and Logistics
Group (www.sclgme.org) has set up a panel of forecasting
experts, with 2 of our experts for a series of webinars and seminars to
address the non-IT issues of forecasting & supply chain planning –
without which forecasting errors will not reduce. The panel has
experience in improving forecasts and service levels in large corporations
(Lipton/Unilever, Reckitt-Benckiser, Panasonic, HP, GSK).
The webinars are FREE OF COST.
If you cannot attend, the minutes can be received and queries can be mailed
to the panel via SCLG after the sign up.
Most companies and software rely on
time series methods for forecasting. However, if average sales variation
is above ~40%, as often, time series algorithms are useless. Only
establishing the cause of variation and incorporating them into forecasting, can
improve accuracy. Sales variations are almost always excess of demand
variation. A major factor for this is the bullwhip effect, peculiar to
supply chains. Frequent changes in sales strategy and supply/stocks
also cause big variations, even while demand remains stable. The
solution lies in:
1. Making demand/inventory information available
upstream (retailer to wholesaler to manufacturer), and make estimates where this
is not possible.
2. Establishing key forecasting parameters in a
a. Impact of price, promotion, sales
effort, stocks, ordering and sales-target systems
a. Tolerance levels, safety stocks,
seasonality indexes... (by SKU/category)
Since the system must be lean (large
planning systems are unwieldy), attendant IT requirement is small. Most
in-house software should lend itself for this.
1. Understanding where time series forecasts work
(& where not) 3. Improving accuracy via incorporating
price, promotion, stocks...
2. Reducing forecast errors by countering the
4. How to manage inaccuracy with safety