Demand forecasting is a well-established and important field in retail, but accuracy and speed are becoming challenges in a fast-changing and ever evolving market. Category managers need advanced support to stay ahead in modern retail.

This is according to Linda Mandyu, Client Success Manager at the Argility Technology Group, specialist retail predictive and prescriptive analytics business within global enterprise, Smollan.

Mandyu says most category managers still attempt to manage demand forecasting on complex Excel workbooks, while some trailblazers are trying to push the envelope by integrating statistical analysis and BI tools. “It remains onerous and highly specialised, and requires very skilled statistical analytics, commercial and retail subject matter expert teams. These skills are hard to find, and when you do find them, they are generally experienced with industry and vertical-specific tools,” he says. “You might need a team of ‘unicorns’ to deliver what modern retail needs.”

As a specialist in the retail sector, Argility combines highly mathematical and statistical expertise with the most advanced predictive and prescriptive modelling, as well as leveraging modern cloud technologies, complete with a skills pool with over 40 years of retail experience.

“We’ve taken that deep experience, combined it with scientific, technical skills and built a product that is fit for the purpose of demand forecasting directly on the world-class Google Cloud platform, with the ability to analyse petabyte-scale data. Then we apply machine learning tools and AI to supercharge those capabilities,” he says.

Mandyu confirms that it is essentially “demand forecasting as a service” that allows retailers to, firstly, improve stock holding and distribution, and then to support price optimisation initiatives, thereby driving an increase sales and retention of your much-valued customers. “It’s like a secret weapon for category and brand managers alike, that arms them with highly valuable intelligence to inform precise decision-making.”

Mandyu explains that Argility’s Predict Retail services deep dive into enterprise data to uncover valuable insights. “For example, if they want to grow market share in a particular category, we look at what competitors have done, matched with their product-price elasticity boundaries, then give them high-level intelligence on where they are over or under priced. We also focus on recommended order levels so they are not over or understocked and can optimise their supply chains and better meet customer demand pre-emptively. This gives value out of the box much faster. It also frees up their skilled resources to do higher value human-specialised analysis that may fall out of the capability of artificial intelligence or machine learning efforts. They don’t have to do the number crunching, but rather pass it to a machine to do the heavy lifting of the incredibly large data sets stretching back over longer time series.”

Mandyu cites the example of an FMCG and perishable goods field service, which is expected to reduce millions of rands in returns through improved demand forecasting capabilities. “While the client closely manages stock holding and distribution and checks shelves several times a day, the process is flawed because it is based on human estimation. The business was incurring over R100 million in product returns each year. We have recommended improved order values to narrow the gap between ordering too much and ordering ‘just right’. The customer is now on track to achieve our conservative target, which is to save anywhere from 2% to 5% in returns, which in this case will amount to millions saved,” he says.

“Complex retail analytics business problems remain a source of endless excitement to the ATG team, a fact that may often seem intractable to our clients. But solving these problems present a perfect opportunity for converging the elusive ‘man and machine’ duo to achieve impactful success exponentially faster, and more accurately, from data-driven initiatives,” concludes Mandyu.