Optimisation is a word much beloved of consultants and IT vendors. We started with ‘price optimisation’ – the ability to stretch price elasticity to the limit to maximise return. Then there was ‘promotional optimisation’ – a preventative art to stop your cut-price offers cannibalising full-price sales. Now there is ‘demand chain optimisation’, which seems to be preoccupied with building consumer demand into the supply chains.
Nothing new here, one may say: companies in the FMCG sector have, after all, been talking about consumer ‘pull’ rather than manufacturing ‘push’ for decades. But that is just the start. Thanks to powerful new number-crunching systems, demand optimisation blends together not only the conventional demand forecast, but integrates every decision along the retail supply and demand chains, analysing the impact on profit and performance of all the options at each stage. Instead of optimistic guesstimates, there is now what AMR Research terms a ‘cost-optimal decision at every point of the demand and supply chain’.
Far more parameters than the average human could cope with can now be added and modelled for optimum performance. It is not only a matter of how many green raincoats you’re likely to sell, but how many can be fitted into the available display space and stockroom; what other sales could be affected; what will be the impact on space, stock and margin if they are advertised on TV rather than press; what happens to the bottom line if the price of green raincoats is pushed up to encourage sales of cheaper sky-blue ones which have a better margin?
Demand-based management
‘Demand-based management is now an achievable target,’ says Greg Girard, vice president, retail application strategies at AMR Research, ‘thanks to the increase in processing power available. Information flows are now more important than product flows in creating and responding to customer demand.’
Many of the software companies specialising in this area are comparatively recent start-ups such as KhiMetrics and Knowledge Support Systems (KSS): ‘We’ve been working in this area for around two years,’ says KhiMetrics’ chief executive officer Brent Lippman, ‘and there has been a lot of educating to do. Retailers have spent a lot of time learning to manage their costs over the past 10 years – but you don’t create great companies by controlling costs, you do it by controlling margin.’
It is also an area the key players are being quick to exploit: IBM is already working with both DemandTec – another specialist in this area – and KhiMetrics, while NCR’s Teradata division has launched a sophisticated demand chain solution.
Teradata has added historic data on promotional media into the modelling mix to improve demand optimisation. The new system would allow the merchandiser to run a ‘what if’ model to assess the impact of, say, two for the price of one pasta sauces. The system would evaluate the impact on related sales (such as pasta) assessing likely upturn in demand and calculating how successful – or otherwise – the promotion could be if pasta stocks ran out or if the merchandiser over-bought to compensate. It can add in the effect of using different media to calculate likely take-up of the promotion and then feed all this back into to the supply and replenishment model to identify potential problem areas.
Early users include BauMax, an Austrian home improvements chain, while The Warehouse, a value retailer in New Zealand, has been using the initial demand management tool for almost two years with significant savings in cutting safety stocks by more accurately assessing true customer demand. One department cut its number of weeks cover by 13 per cent with no loss in sales, while the toy section was able to reduce inventory by five per cent while sales rose by 20 per cent by using more accurate forecasts.
It has taken retailers some time to appreciate the benefits of these sorts of tools, but as Brent Lippman notes, in the past year the picture has changed dramatically: ‘As price, promotions and replenishment come together to create the new demand chain model – and with many users in the US reporting payback within months – it may not take too long for this latest ‘optimisation’ technique to gain ground.’
