Talking to retailers in Australia about Artificial Intelligence, it doesn’t take long for the conversation to turn to the impact AI is having on demand and supply planning. After all, the goal of planning is to mitigate working capital requirements against stock availability aspirations and that’s a problem AI is perfectly for.

As a demand planning system, Quantiful is proven to have a positive impact on both demand and supply planning. It improves cash efficiency, revenue and margin. However, the benefits of bottom-up SKU location planning (which underpins Quantiful’s approach to demand planning) are proving to go well beyond improving the basic financial metrics. 

In this paper we describe 5 les-expected outcomes Quantiful customers are benefiting from as a result of deploying Quantiful at scale inside their operations. They are far from exhaustive, but they should go some way to demonstrating both the breadth and the depth of impact AI can have on retail.

• Optimise assortments at scale

The days of ABCXYZ planning for inventory and assortment are fast becoming outdated. To this point, this rather crude method of store categorisation has been widely used to determine the assortment depth and breadth but it fails to take into account the ever growing store-level nuances that retailers are dealing with.

The most successful retailers have store assortments that precisely meet the needs of the consumers that shop there. For this to happen, we need to let the demand of the consumers define what product and how much of that product is available in particular stores. 

With AI, retailers can build demand models that optimise store space (GMROF) with the products that we know are going to sell the most. They run dynamic stock levels that are based on SKU Location forecasts. For many this is a significant shift - but an essential one, as it allows them to dynamically set the assortment by catchment and remove the need to cluster or categorise based on legacy methodologies. The result is significant improvements in customer experience along with bottom line profitability.

By understanding the needs of a store’s specific demand profile Quantiful can dynamically support the assortment planning for each retail location. This typically results in a 65% reduction in lost sales, 10% increase in-store revenue, and 18% decrease in the store SKU count, or more more importantly dynamically setting assortments to meet the catchment demands of the store.

• Track business performance at individual store level

Given the limitations of traditional reporting systems, most retailers have little choice but to look at a ‘whole of business’ view of their operations. Without the ability to look at granular store-level performance they are missing out on the ability to see a myriad of store-level opportunities to make incremental improvements to assortment, stock and discounting.

In the same way AI allows SKU location forecasting, it can also be used to pinpoint areas of inefficiency at a store level. With SKU location data, retailers can, for the first time, benchmark individual stores using metrics that were previously reserved for whole-of-business analysis such as CCC, GMROI, GMROF, availability and turn. Similar to identifying poor performing products, the cash-use efficiency from the bottom up can be used to identify poor performing stores.

With bottom up data, Quantiful customers can benchmark individual stores using common supply chain metrics and build store level business improvement plans that drive more sales, better margins and improve stock turn.

• Run limitless ‘What if?’ Scenarios

Being able to run relatively complex ‘what if’ analysis is a luxury very few retailers have. Without a powerful analytics engine, a mountain of data and some super smart people, retailers too often rely on gut to make important decisions - and often the decisions they make are sub-optimal. 

The benefit of good SaaS products which leverage AI is scenario planning will execute the entire portfolio forecast independently on all products. This means understanding the cannibalisation, substitutional or even the halo effect one decision has on another product line, and all executed in virtual real time.

For example, AI has the horsepower to understand:

• The relative variable influencers of demand on SKUs, this could be discount sensitivities, weather, date etc. This means a retailer can predict, with accuracy, the effect of a price change on a particular SKU in a particular store will have on other products in that store, or in fact other stores such as e-commerce
• Being able to scenario plan regionally allows maximising contribution margin based on the demand or behaviour of the catchment area the store is serving

These two capabilities  allow business to do powerful scenario planning - answering questions like

“If I wanted to win market share in the appliance sector in New South Wales, what price point would I need to meet, and what impact would it have on the sale of vacuum cleaners?”

“If I needed to mitigate obsolescence in a particular product, where in Australia should I send it in order to sell it at the highest margin and mitigate the on the sale of other products?”

Even with a high SKU Location count, Quantiful allows planners to scenario plan at any level in the product hierarchy, or any level in the distribution network with confidence the SKU Location execution will be of 75% accurate resulting in fast agreed trade marketing executions

• Change-up store work

As a retailer, you want your in-store teams focused on selling, engaging with customers, merchandising and making sure the store looks appealing - all the things that make the till ring. What you don’t want is your store team doing admin and fighting with stock issues (too little and/or too much).

The real pain associated with poor short-term planning can be seen on the shop floor. To keep shelves stocked, managers need to execute manual orders, ad hoc receipting and assortment planning which all take valuable time, time that should be spent selling! 

With Quantiful driving the planning process, store managers create 75% fewer manual orders and out of cycle receipting, which frees them up to focus on customer and revenue-related activities.

• Get planners back to planning

The introduction of a centralised planning system inevitably has an impact on the day-to-day work that demand planners do, particularly if they are coming from a world of spreadsheets.

The complexities of today's portfolios and distribution networks means Excel has become onerous to manage, it's slow, unfit for purpose and more often than not, sits on an individual's laptop. We see too many planners spending far too much time trying to reconcile top-down forecasts into tactical operational outcomes.

With more accurate planning at the front end, the number of forecasts that have to be manually adjusted drops precipitously. Quantiful customers have found they only have to manually adjust 2% of forecasts at store assortment level (compared to over 60% previously) when 98% of forecasts are running unsupervised actions. With more time to think, planners can get back to what they are good at: supporting sales generation across an agreed plan.

Quantiful is generating over 50 million unsupervised replenishment actions for retail customers across AUS/NZ, resulting in an average of 35% working capital requirement and 55% improvement in SKU Location stock turn (by revenue)