Scaling a modern retail business requires a combination of automation and predictive analytics. Here’s how to leverage these technologies to get real results.
Across nearly every industry, business leaders are touting the benefits of automation — and there are plenty. Automating certain repeatable processes within a business is the key to freeing up human workers to do the “big-picture” strategy work required to scale up.
However, simple automation on its own is often not enough, especially for specialized or highly dynamic retailers. There are too many exceptions — different business rules, preferences, store types, product types, vendor and shopper behaviors — to see consistent results from automation alone. More importantly, automating inaccurate decisions will just streamline mistakes, resulting in lost sales again and again.
To truly make a difference for your retail business, automation should go hand in hand with AI and predictive analytics. Using these technologies together allows retailers to scale their business to compete with the Amazons of the retail world, while increasing profit margins and maximizing revenue.
Relying Solely On Past Data Doesn’t Cut It
Looking only at historical sales for automated processes is generally unreliable for a few different reasons. It doesn’t account for:
- New products with little or no history
- New stores or sales channels
- Changes in competition
- Last year’s promotions that won’t happen this year
- Moving holidays like Easter
- Changes in vendor costs and lead times
As an example, let’s say you only sold 150 laptops because you ran out of stock, but there was enough demand to have sold 250 laptops. If you base future demand solely on past sales numbers, you will only re-order 150 laptops, essentially repeating this mistake in the future.
Forecasting demand manually is time-consuming and inefficient, but using automated methods that don’t account for dozens of critical factors and solely rely on past data is a huge mistake too. By combining automation with predictive analytics, you can analyze past sales in a more intelligent way to generate more reliable forecasts.
Retailers need to let predictive analytics and AI engines analyze their past sales first before automating future decisions, to account for dozens of influencing factors that can create a skewed perspective on past demand and will affect future demand.
How To Use Automation With Predictive Analytics And AI In Your Retail Business
There are several key operational processes that can be vastly improved with the right combination of automation and predictive analytics:
1. Inventory management
Effective inventory management requires you to accurately anticipate customer demand, vendor performance, what future inventory levels will be like when your new orders arrive and more.
A smart predictive analytics system will generate suggested order quantities to bring the right products to the right store at the right time. You’ll be able to generate open-to-buys based on merchandise and assortment plans, while accounting for many business rules, vendor constraints and future demand.
Predictive analytics also can help solve some common challenges among specialty retailers, such as size distribution in fashion retail. The most common sizes (small, medium and large) are often sold out, while the retailer is carrying XS or XXL in excess.
Retailers are often challenged to determine an optimal size distribution, since different products and styles will experience different consumer demand across different locations. This variance in customer behavior, as well as differences in geo-demographic attributes, creates a “size curve” that is unique to each product.
Calculating this size curve for each product, especially in the highly seasonal fashion industry, is mathematically challenging and time consuming. So retailers end up generalizing, basing size distribution on past sales or similar items, and marking down the inventory that doesn’t sell.
For example, if you have a store that tends to pull in taller people, your size distribution will be skewed towards the taller sizes, which results in more large and X-large sizes sold at this store than the smaller sizes. If this curve isn’t considered throughout the inventory management process, then this retailer will be left with a lot of smaller sizes that no one bought and will also run out of stock on the larger sizes that are in high demand. However, with predictive analytics, they will be able to see specific trends like this for each product and make more informed inventory decisions, as well as automate this decision-making process in purchasing and replenishment.
2. Price management
Predictive analytics and artificial intelligence are able to optimize prices throughout the entire lifecycle of a product, from initial introductory prices to regular prices, to promotional prices and finally, markdown prices.
The price management process requires a business rule-based system that answers the following questions:
- When does a product’s price change?
- How much should it change by?
- How do competition and vendor costs trigger changes?
- What is the targeted gross margin on the product?
- How do other related products need to change in price?
- How do the prices compare in different store types or geographical price zones?
With a predictive analytics solution that operates within the constraints and rules mentioned above, you can mathematically calculate the optimal price for any given product.
Technology also can help when it comes time to marking down a product being phased out of inventory. Most retailers just arbitrarily mark down their products in a last effort to get rid of the merchandise. In many cases, retailers are making no margins on markdowns, they are just trying to clear their shelves and recoup costs. Even retail giants have serious overstock problems, to the tune of $4 billion, in the case of H&M.
Predictive analytics starts working proactively on an optimal markdown strategy. It accounts for which products you have an excess supply of, how quickly you want to get rid of them and the best pricing strategy to get you there while maximizing gross margin.
Retailers run thousands of promotions every year. Although in many cases parts of this process may already be automated, the problem is that they run them often without any advanced analytics tied to inventory levels at the stores, demand or other critical factors.
Here’s just one example of how promotions can lead to unintended overstock: If there are two similar products (item A and B), and a retailer puts item A on promotion, demand for it will increase because of the lower price. The customers that were going to buy item B are now going to purchase item A because it’s cheaper. This will lower demand on B below initial estimates, leaving it overstocked at the end of the season.
This effect — promotional cannibalization — can be avoided with predictive analytics and AI. Similar to a proactive markdown strategy based on analytics, retailers can understand the potential impact of a promotion on other products before running it and weigh the pros and cons of doing so. You may end up learning that it’s better to promote a different product to balance out demand, thereby avoiding unintentional promotional cannibalization.
Analyze First, Then Automate For Retail Success
There’s no question that automation technology is a wise investment for any retailer, and it does have the potential to transform your operations. Instead of simply automating a process that will replicate last year’s mistakes and miss new opportunities, introducing predictive analytics and AI into the process will not only save time and resources, but more importantly increase margins and maximize revenue.
Yan Krupnik is the Director of Business Development at Retalon, an award-winning provider of retail predictive analytics solutions for planning, inventory management, merchandising, pricing and promotions. Retalon’s solutions are built on one unified platform to account for all factors influencing your business.