As customer acquisition costs and competition increases, creating new revenue opportunities with existing customers is fast becoming the most cost-effective way to grow a retail business. Coinciding with this is an increase in the number of SKUs that retailers carry. As acquisition costs and SKUs continue to grow, it’s vital that retailers take a hard look at sales data to understand what they can sell more of and how.
One such way is market basket analysis. Market basket analysis uses point-of-sales data to uncover products that are frequently purchased together (aka product pairings) within a retail setting. Once they identify statistically significant pairings, retailers can potentially maximize sales of both products.
In establishing product pairings outlined above, market basket analysis gives retailers valuable cross-selling and promotional opportunities. These opportunities can grow average basket size and introduce customers to new or related products that they may not have been aware of.
In this post, we’ll shine a light on the arguments for and against market basket analysis in brick-and-mortar retail.
What are the Benefits of Market Basket Analysis?
There are a lot of areas of your business that can benefit from market basket analysis.
Capitalizing on customer purchasing behavior
By having all of your sales data stored and centralized, you can start to draw relationships between products that end up being purchased in the same baskets. By analyzing sales data, retailers may more effectively cross-sell, make targeted product recommendations, and reap maximum rewards from marketing campaigns and promos.
As an example, a Walmart Superstore stocks 142,000 different items in a single store. The amount of pairings possible with that number of products is staggering. Not to mention, the volume and frequency of these pairings over just the course of a single day. To illustrate the point, if you form a set of three from a data set of 100 items then 161,700 combinations are possible.
But retailers can’t go chasing after every three-product set that occurs in some frequency in their stores. If, however, you know which combinations are the most sellable and provide you the most revenue, you can prioritize and promote those. By bringing these high-frequency combinations to light in your shoppers’ minds, you’ll get the most out of market basket analysis, plus some deeper insights into your customers’ purchasing preferences.
Fresh visual merchandising
Thanks to market basket analysis, you no longer have to guess what is or isn’t working. By combining your store’s heat map technology to see what’s getting attention with your market basket analysis, you can merchandize and market in such a way that encourages cross-selling of frequently paired items.
An example of this would be the discovery, via market basket analysis, that customers in your c-store who buy a small sandwich typically also purchase a dessert of some kind. Thus, you could place visual merchandising for small sandwiches near the candy/cookie aisle so as to encourage cross-selling.
Market basket analysis gives you the ability to pinpoint your merchandising towards verified upsell and cross-sell opportunities. This insight, while not perfect, does remove a lot of the guesswork that comes with understanding how and what to merchandise.
Supply chain efficiency and demand forecasting
If you know how often and how much product will sell in a specific market, you can better optimize your inventory… at least theoretically.
If you know what shoppers are likely to purchase when they buy other products, then you might consider not only merchandising them together on the floor, but in your stockrooms and warehouses as well. In a world where speed and convenience are king, it makes sense to organize your stock in a way that maximizes efficiency. If retailers can reduce the time it takes to pick items that are purchased together, you can get the most out of your labor spend and get products to waiting customers sooner.
From a timing perspective, market basket analysis sheds light on what times of day, week, or year product combinations are more likely to take place. For instance, perhaps pre-prepared dinners are likely to be purchased alongside soft drinks or beer in the evenings, versus cleaning supplies and paper towels on a Sunday afternoon. This predictive insight ensures that you’re not left on empty at crucial buying periods for certain products.
What are the Limitations of Market Basket Analysis?
Market basket analysis on it’s own will still leave room for improvement.
Averages tend to lie. If you’re trying to duplicate a conclusion drawn on chainwide data to merchandise a single store, you’ll hit some speed bumps. An infamous version of this pitfall in the retail world is that of the ‘beer and diapers’ correlation. Way back before ‘big data’ (the ‘90’s), a retail company ran SQL queries against its store data and discovered that beer was often purchased with diapers. This discovery quickly caught wind, and stores started to put diapers and beer nearby on store shelves. Naturally, sales of the two together went up.
The issue is this… Of course sales went up for the combination of items; stores were merchandising them next to one another in a highly trafficked area. The root of the problem is that you can’t seek out and validate correlations in data that you had a hand in creating. Thus, although market basket analysis may help you spot a trend, once you act on it, it’s difficult to assess the validity of the correlation.
No clear calls to action.
Even if there’s a true link between the two products, it takes time and skill to figure out how to act on this information. For example, let’s say you learn that 50% of customers who buy bread buy rice within two weeks.
You could use your market basket analysis tool to have a recommendation sent to each online shopper who bought bread within two weeks of their purchase. But when do you send it to ensure you capitalize on the correlation? The tool can’t tell you. And how can you use this insight in your physical stores? In a best case scenario, your customer has an app or loyalty card that lets you track their purchases. With that, you could send an app-based promotion for the rice. But given that most Americans are enrolled in an average of 29 loyalty programs at once, the odds of your message being received is low.
Test and learn lag times.
Because there are no clear calls to action, retailers who use these solutions must dedicate time to A/B test any actions they take as a result of the data. But how do you decide which correlation to A/B test? Testing even just the strongest correlations takes enormous time and effort.
Let’s say you manage to pick a product pairing that you think will generate the most revenue. There’s a lot of work to make the test happen. For instance, if you’re going to run an in-store cross-promotion, you’ll need to re-organize your shelves, rework your planograms, and send directives down to all stores. From there, you’ll need to train staff on the new locations and make them aware of the promotion. You’ll then need an adequate amount of time (weeks? months?) to test whether or not you were right.
Costs aside, from learning the correlations to making changes and then testing their efficacy and using what you learned moving forward, this entire process is slow.
The Big Picture
Market basket analysis gives insight into product relationships at your store that you may never have known existed. Using tech that analyzes market baskets can reveal novel ways to cross-sell, up-sell, and promote your merchandise. Given that acquiring a new customer can be 25 times more expensive than retaining an existing one, retailers are eager to find ways to increase average basket size and drive customer lifetime value.
These opportunities aren’t without costs, however. You’ll need to assess your store’s priorities before deciding to implement changes based on the insights offered by market basket analysis. After all, market basket analysis can’t do a lot of good if you haven’t settled on an overall pricing strategy or aren’t tracking customers with a loyalty program. And physical stores are limited when it comes to ways to capitalize on insights gains.
Sometimes CB4 is confused with market basket analysis. That’s because, like market basket analysis, CB4 uses transactional data for multiple products in tandem to help retailers generate new revenue. The difference is that CB4 isn’t spotlighting high-revenue pairings. We show our customers high-revenue, in-demand products that aren’t meeting their potential now in a store or stores. Then, we guide store managers on how to capture opportunities to sell more with quick, easy fixes.
Sound interesting? Learn more about how the solution works in this short video.