"I am passionate about the SIOP process, worked hard to put it in, but until we had that credible forecast, I couldn't start that process," Lewis said.

So, Lewis said, the planning changes had both a direct cost reduction component relative to inventory - he knew the investment would more than pay for itself - as well as more strategic side in terms of enabling an SIOP program and driving increases in the top line through better new product introduction.

A key capability of next-generation planning tools, Lewis said, is the use of "probabilistic" forecasting technology, a new approach versus the traditional "deterministic" forecasts most companies rely on.

How does this work? Lewis noted an example of an unexpectedly large order that might be received by RS.

"We were really interested in a tool that would tell us 'You know that very large order that you just took, well as nice as it is for the business, it actually has a very low probability of recurring, and that means we're not going to massively increase our forecast in the future.' " Lewis said.

He noted that in the past, such large orders often led to big jumps in future forecasts, and subsequent costly jumps in inventory levels for the SKUs involved.

RS Wanted a "Black Box" Forecasting Approach

Many companies want to move from a very hands-on approach to forecasting, where the system handles the forecasts for most SKUs and demand planners deal primarily with exceptions , or SKUs for which something seems out of whack.

Given its huge number of SKUs, RS needed to move beyondeven that level to more significant levels of automation of the forecasting process. Lewis said RS had reached a level where demand planners were in part measured by how many forecasting exceptions that they could clear each day. That even as exception management was one of the least favorite tasks for many planners, and which was among the activities that produced the least "value add."

That higher level of automation also enabled Lewis to make changes to how his demand planning team, some 200 people in total, was structured.

"For some time now, I've wanted to split off the initial creation of the statistical forecast from the subsequent enrichment of it by the demand planners," Lewis said. "I want my demand planners to be people people - I want demand planners to be able to bring out from other functions things the system can't properly know."

Another important capability related to being able to build "service classes," or the ability to take like groups of products by different attributes and set service levels and safety stock policies unique to each group.

Lewis said that has proven very valuable for the management of new product introductions (i.e., modeling a given new product on the history of a similar product), and also for grouping different products together for certain seasonal or event-based requirements.

"If there is a flood, for example, we want to be able to associated the high visibility vests that will be in demand with the sand bags, with the signs, with the torches, with the batteries, etc.," Lewis said, noting that these products may actually be in several different planning buckets individually.

He also said RS has achieved great results said with regard to NPI through the ability of the new tool to more quickly perceive "rising stars," or new products that are seeing high levels of demand. Now, RS can more quickly and automatically get the right levels of supporting inventories to meet the demand for those SKUs, helping to improve top line sales for the company.

Benefits are Very Real

"When I approach our CFO and said I want to put this tool in, thinking the full costs to put this in are going to be about half a million pounds, we hadn't budgeted that number. So I was going to the CFO with no budget, saying can you give me half a million pounds please," Lewis said. "I told him if he gave me that money, I would give him more than 2 million pounds back still that financial year [more than halfway through the year], and several million more the following financial year."

Amazingly, Lewis said they save those 2 million pounds in the first three days the tool was live - with the new safety stock calculations, purchase orders exceeding that level were canceled just days after turning the system on.

"That stock has stayed out of our system ever since, and as the recovery started, it allowed us to be much more selective about where to add stock back in," Lewis added.

The new system and approach a number of benefits. Lewis said, for example, it helped identify high volume SKUs for which planners tended to put in lots of safety stock to ensure high service levels. But in fact many of those SKUs had very responsive suppliers that could carry part of the inventory load.

Lewis said that in some cases RS has been able to reduce inventories as much as an incredible 80% on these fast moving SKUs, either freeing up cash for the overall business or enabling it reinvesting some of the inventory savings in other SKUs with more volatile demand.

With this great success, the ToolsGroup software is now being rolled out in Asia and North America, Lewis says.

Interesting Lewis also believes that companies should more aggressive in terms of adopting supply chain technologies.

"I want to treat this type of supporting planning technology like a mobile phone. I want the latest and greatest, and I want the best of breed," Lewis said. "These tools can pay for themselves very quickly if you are rigorous about implementation...I'm very happy to unplug one tool and plug another one in."

What is your reaction to this RS Components case study? Do you think there is a need for "next generation" planning tools that are more automated, and using more advanced technologies such as probabilistic forecasting? Let us know your thoughts in the Feedback section below.