How to improve demand forecasting on short shelf-life products


Retail

Over the bank holiday weekend, I went to a large supermarket chain to buy supplies for a last-minute BBQ only to find they’d run out of burgers and buns and most importantly, charcoal. I didn’t have the energy to spend the last few precious hours of my bank holiday traipsing around supermarkets, so I quickly gave up and begrudgingly had a jam butty for tea.

But in today’s day and age, with the power of big data and predictive analytics, no food retailers should ever suffer loses in sales as a result of running out of products – perishable or not.

We can’t deny the globalisation of supply networks makes the task of supply chain management more and more challenging for food retailers and it often requires strategic shifts to continue to meet market demands.  But without proper demand forecasting processes in place, it can be nearly impossible to have the right amount of stock on hand at any given time. In the case of perishable food produce, an unexpected dip in demand will result in waste and loss whereas a spike leaves orders unfulfilled and consumers buying elsewhere (or in my case, not buying anything at all). But when predictive analytics is combined with machine learning it can give a more accurate forecast than is humanly possible, minimising cost and waste and maximising margins. By forecasting demand accurately, food manufacturers can better plan production, distribution, storage and supplies when they can get answers to the following questions:

  • Will this product demand be influenced by external factors like weather or holidays?
  • Does this product follow classic seasonal patterns?
  • Is it impacted by geography (demographic and social trends, regional preferences, macroeconomic environment?)
  • Can it be identified as seasonal riser/decliner or sustained riser/decliner (growing year over year or sudden growth within the past months)?
  • What consumer behaviours are trending that may have an impact?
  • What new or up-and-coming technology trends will impact demand (the increasing use of third-party mobile food delivery and digital coupons, for example)?
  • How is the product marketed to consumers compared to the industry standard (price-point driven for value consumers, or appeals to another segmented market)?
  • What will the consumer demand for a new or limited time offer product be?
  • What similar products are your competitors developing/launching or promoting and how will this impact your product?

Data collated from sales history, weather, special events, promotions and competitor activity (as well as the above) can be used to forecast what consumers will buy, how much and when. This data then helps food manufacturers to plan what to make, when to make it, what supplies to source and when they are needed, leading of course to happy customers that never go without a BBQ on the hottest August bank holiday on record (I’m still quite bitter about it).

The other benefit of demand forecasting is of course, correctly pricing your products at a cost lower than other retailers, but in a way that still gives you maximum profitability. Customer’s expect the food they eat to be affordable. Price something too high and you’ll lose custom to your competitors, price it too low and you’ll reduce profitability.

Many food retailers are currently “price followers,” making reactive changes based on fluctuations in commodity prices, or competitors’ price shifts. This approach can inevitably lead to a chaotic pricing architecture across products – but no one wants their bread fluctuating in price every week.

A demand model capable of predicting volume changes and prices in of any product (including the competition) could be created to analyse each brand to determine the consumers’ sensitivity to price shifts. These findings can then be fed into an optimisation model to determine the right pricing actions for a food retailer moving forwards. This model can create the opportunity to estimate the best course of action under a variety of market scenarios meaning your customers are not only able to buy the products what they want when they want it, but you’re also able to keep costs down, maintain margins and keep your customers coming back for more. Win-win!

To find out how you can improve demand forecasting and achieve meaningful outcomes, faster, check out our Data Platform Accelerator.

Posted by Helen Thomas