Optimizing Online Retail Industry using Artificial Intelligence: Leveraging Relex Software Solutions Results for Retail

Comments · 2200 Views

RELEX Solutions is dedicated to helping retail businesses improve their competitiveness through localized assortments, profitable use of retail space, accurate forecasting and replenishment, and optimized workforce planning. Their SaaS solutions deliver a quick return on invest

Online stores whether for a pure e-commerce retailer or as part of an omnichannel operation allow retailers to offer customers a vast assortment that simply can’t be rivalled by traditional brick and mortar locations. Online retailers aren’t even physically limited by warehouse space, as not all products need to be stored in their warehouses. It’s quite common for retailers to place supplier orders for most items in their online assortment only when a customer makes a purchase. This level of flexibility allows retailers to boost online catalogues that run into the millions of items, even if most are never sold. 

Automating E-Commerce Assortment Decisions

While traditional inventory limits may not apply to many online retailers, they still need to make smart, data-driven decisions about what to stock if they want to attract and retain a loyal customer base. E-commerce inventory management should never be approached as a one-time or even as a periodic decision. The pace of online retail is simply too fast for that.

While retail, in general, has always been rich in data, online retail, in particular, provides an absolute abundance of information like page visits, visit durations, click-through rates to other products, conversation rates, and more. More data means more opportunity to curate the right assortment. Online retailers should be constantly monitoring their stock and data, making sure their assortment hasn’t gone obsolete in the eyes of their shoppers.

A data-driven model like this can provide a reliable foundation for the automation of inventory decisions. Retailers who invest in technology that can accurately automate these decisions ultimately free their inventory managers to focus on higher-level tasks, rather than asking them to create assortment recommendations from scratch on a continuous schedule. Inventory managers can take the data-driven output from their inventory management system as a starting point, then integrate external factors and their expertise as exceptions to the automated recommendations.

Choosing the Right Stocking Strategy

But inventory strategy goes beyond assortment choices alone. Retailers still have to implement stocking strategies that make sense for their business.

After real estate, labour, and all manner of other costs, keeping inventory in your warehouses isn’t cheap. Still, warehousing overhead can be balanced against the fact that unit-costs are generally lower when retailers place large batch-size orders with their suppliers. The amount of time spent per product in all phases of the ordering process is also quite small, and you can take advantage of freight-free limits and cheaper means of transportation.

When ordering against customer orders, on the other hand, the cost of warehousing tends towards zero. However, per-unit costs increase significantly when retailers source in small batches or even single units, as some do. Shipping costs also rise steeply when retailers need to quickly transport purchased items from multiple suppliers to their customer. The whole endeavour can grow quite expensive, even without significant warehouse costs.

At the end of the day, online retailers have to compare the costs of these two models and decide what balance of the two will be most productive and cost-efficient for their business. As a general rule, though, it makes the most sense to stock low-priced items that sell well in your warehouse, and order high-cost items that sell infrequently directly from the supplier’s warehouse.

There are always, of course, several exceptional situations that need to be addressed, most of them concerning sales rates. For example, there’s no way of accurately predicting how well a new item will sell. Still, retailers can start from a baseline estimate that draws from previous introductions of similar products. As data on the newly introduced item populates the inventory management system, it should be capable of automatically updating its decisions. Once this basic configuration is in place in the inventory management system, managers can easily tag exceptional items to ensure they stay in stock.

While the underlying principles at play are simple, gathering and maintaining all the relevant data then running these processes efficiently is a complex challenge. With the right technology strategy, though, online retailers can optimize and automate their inventory management decisions to maximize their profits.

Principle Working 

Choosing the right replenishment days for products and stores is very important for a retailer’s operational efficiency. Although the high delivery volumes in grocery justify daily deliveries, it does not make sense to replenish all products daily. Large, consolidated deliveries of products displayed in the same aisle of a store make in-store replenishment considerably more efficient. As more products can be shelved in one go, store associates do not need to waste time moving roll cages around the store.

To achieve this, the supply chain planning team had defined so-called main replenishment days per store and display group, i.e. products displayed in the same area of a store. This approach had already proven its value, but the store replenishment team struggled to define the main replenishment days manually for hundreds of stores with different floor plans and demand patterns.

RELEX’s AI-based optimization of main replenishment uses a swarm intelligence algorithm for multi-objective optimization. The algorithm prioritizes between objectives based on customer-specific business needs.

In some cases, the main priority can be to attain as smooth a goods flow over the week as possible. In other cases, minimizing deliveries during weekends, when labour is more expensive, can be the main priority. the optimization minimizes the number of main delivery days and expected shelf breaches, The optimization is done per store to find the best main delivery days per groups of products – typically products that are displayed in the same part of a store – taking the products’ sales patterns, shelf-life, and other relevant restrictions into account.

  • They are now able to automatically optimize the main replenishment days with little manual work. The optimization is also more granular, enabling the planners to assign more main replenishment days to top-sellers compared to slow-movers within the same in-store display group of a store.

  • The benefits of applying main replenishment days have increased as AI chooses the replenishment days more accurately. The stores receive more of the same and similar products in one go, making in-store replenishment more efficient, without suffering negative consequences. On-shelf availability, spoilage or the proportion of deliveries fitting straight on the shelf have stayed on the same level or improved, while deliveries have become more consolidated to the chosen main delivery days.  

  • The inbound goods flow to the stores is smoother, resulting in a more balanced workload for store staff, more predictable work shifts, and significantly fewer capacity issues.

Figure 1: This graph shows the inbound goods flow to a hypermarket before and after the main replenishment days had been optimized using AI. As can be seen, the goods flow following optimization is much smoother.

RELEX system automatically provides order proposals based on accurate forecasts – even for seasonal items.

  • Plan its capacity and workforce

  • Handle the purchasing process and optimize purchasing costs

  • Optimize inventory and stock values

  • Work on computing all the seasonal indices and other parameters using principles of operations research and supply chain engineering models like EOQ and materials planning inventory

Put an effective forecasting and replenishment solution in place quickly, one that could handle split deliveries and changes in demand due to promotions and holidays. They also wanted to reduce waste. 

RELEX uses machine learning abilities to include numbers into the forecasts so that companies can use inbound figures to predict demand. The system processes data provided by some input and early indications show that this can help improve sales forecast accuracy.

RELEX’s solution uses over 3,000 forecast model combinations and numerous parameters including historic sales patterns, seasonal effects, supply days, minimum order quantities, freight-free limits etc.

Results

The results are based on various projects taken by RELEX solution and are direct outcomes of the projects. These are solutions to different clientele requirements and are varying in the project out as per the goal described by the client. 

Availability of the most important product lines had improved by 4%, which has boosted sales performance. Increased availability from 93% to 98% across ambient products and in-store availability 85%. At the same time, waste has decreased significantly. Purchase order automation levels were pushed to between 80% and 90% reduction in ordering time. Spoilage was cut by around 40%, while that of chilled products fell over 20%, significant because that category accounts for around 20% of total wastage. 

Forecasting has become more granular which means we get day by day prediction of forecast giving better planning power to the retailers and suppliers to forecast accuracy improved 2–3 weeks.

Inventory has fallen 8–10% (capital cut 3.1%), 20% inventory reduction in key areas targeted by the team.

2.2% increased sales from newly planned and replenished categories.

6-8 % reduction in personnel costs. Improved employee engagement and lighter workload for store managers

Better data management from 91% to 97+%

During the 7 weeks, inventory turnover increased by 133%, eliminating slow-moving stock and creating space for new titles.

Case Galexis - Achievements

 

Conclusion

For any online retail platform or even brick and mortar facility inventory is the most important form of resource. Optimised inventory levels reduce cost significantly on an average for an E-commerce SKU

The cost associated with inventory = 62$ 

so if you reduce it by 8%-10% you get a figure of 56$ per SKU.

Profit associated with a medium-sized product = 110$ (average profit margin calculated from an E-commerce platform keeping a profit margin of 20%)  per product

If this software improves availability by say 4% at an average as per the results given previously

per 100 products we gain a profit increase of 440$.

Usually in E-commerce companies availability matters when a product is high selling or during offer periods for which demand planning and forecasting needs to be very accurate. With the help of this AI software, we will be able to generate forecasts with seasonal indices and other parameters. This will ensure better availability and optimised inventory levels.

Another potential loss of business is customer satisfaction whose cost is estimated in terms or reorder or recurring visits to the platform. The approximate cost associated is again measured in terms of the average profit associated with medium-sized products.

6% of the customers come back to shop at the same platform

2% of them are reorderedEcommerce-conversion-rate

Fireclick global conversion rate

Say 100 customers shop once and like the experience out of which 2 reorder the same product giving 220 $ profit margin and 6 repeat customers buy 1 product each again giving an average profit margin of 660 $

Another aspect of having an AI SaaS is to save money on the manual computing cost.

The average cost for high skilled labour in the supply chain industry = 35$/hr

With the above stats, one saves 8% in personnel costs which gives a gain of approximately 3$/hr

This also saves labour from doing manual and lethargic work thus allowing them to focus on process improvement and continuous improvement activities.

References

https://www.invespcro.com/blog/the-average-website-conversion-rate-by-industry/

https://www.relexsolutions.com/

https://corporatefinanceinstitute.com/resources/knowledge/accounting/product-costs/

https://www.feedspot.com/#all/all

 

Comments