How Market Basket Analysis Can Skyrocket Your Product Sales

With the ever-changing retail landscape and dynamic market, retailers are always thinking of improving their sales. Whether experiencing a revenue slump or adjusting to the new store location, even a slight dip in sales already sets off the income alarm.

And if you’ve managed to hit your targets, there’s always room for more, or at least, you want to sustain the momentum. It leads you to double your efforts in promotions raising your customer acquisition costs. As such, many would upsell and cross-sell to their current clients. However, some of them will experience brand fatigue or move on to other outside offers.

So, how do you combat the challenging times? How do you attract new clients and reactivate your customer database? Better yet, how can you increase your sales? Of course, there are still ways; one is using market basket analysis.

Goal of market basket analysis

Market basket analysis uses mined data that shows what a customer buys, their reason for doing so, and what other products they buy it with. The term itself is based on the idea of a shopping cart, or basket if you will. MBA provides a descriptive analysis and unlocks any associations across products and services. This way, we understand what drives every purchase, which allows marketers to direct their promotions on the bestsellers and the consequent products.

Understanding market basis analysis

Market basket analysis is a method that retail managers use to identify purchase patterns involving two or more products based on their affinity. It has been observed that consumers have a predisposition to buy a second product, a third even, alongside one another, like they can’t get one without the other. Or one product sets off the other, but interestingly, not interchangeably so. Such a relationship across products should be considered with your promotional strategies when looking for a sales boost or overall improvement of your revenue.

A perfect example of applying market basket analysis in a food business is people’s preference for burgers and fries. When you go to a fast-food chain restaurant and decide to dine in, it’s almost impossible to buy a burger without fries, and a drink, too. Hence, we see value meal sets in the menu with built-in discounts because they capitalize on your affinity for getting these products together. It’s also worthy of mentioning that because of the deal price, even if somebody just wanted an ala carte burger to go, they’d be foolish not to get a meal instead. After all, it will only cost just a little to add more value to the purchase.

Ok, maybe burgers and fries are pretty obvious. As such, it is where a more comprehensive market basket analysis comes into play to determine correlations across products and services. We need to look for certain attributes that trigger the purchase of products in combination with one another, whether one acts as the antecedent or as the consequent. As we mentioned earlier, it’s not interchangeable such that A = B, but not B = A. You can optimize your marketing by focusing on these items and adjusting your margins for maximum profitability when you know which is which.

How to apply market basket analysis

When going into the restaurant business, insights gained from market basket analysis of a similar business model can help you formulate a menu and promote the items in it. As such, you can use these product combinations, or item sets whether you present them separately or as a bundled offer. However, you will find yourself working with a large volume of data during research for more accuracy and reliability. Hence you will need to use computational statistics to arrive at the item sets shown on an if-then statement.

A = B


1. If ( burger), Then (fries)


2. If ( burger, fries), Then (sundae)

In the first example, a burger is considered a primary product (antecedent), one that a customer will think about buying first. And because they are buying a burger, the equation tells us that they are likely to buy fries (consequent). The same goes for the second example, which shows two items or an item set that causes the purchase of the third item. However, when somebody just wanted fries, it’s not likely they would want a burger.

Once you’ve established this relationship through market basket analysis, you can also vary your margins based on where the items are placed in the equation.

  • Point-of-sale transactional data such as the company products and/ or services
  • Rules in the form of an If-then statement

Association elements of market basket analysis

Three measurements explain relationships or associations among products that make up rules: support, confidence, and lift. For example, if burger, then fries, we want to know how good a rule it is, which will guide us in our pricing and marketing actions.

Support is the proportion of transactions involving the items in your rule. In other words, you want to know the odds of a customer buying the items in the first place. You can compute this manually by counting the number of transactions with these items and dividing them by the total number of transactions. For example, there are 300 purchase transactions by a fast-food chain in an hour. One hundred of them included a burger and fries, which puts its support value at 0.3. A high support value also helps you identify your keystone products or bestsellers.

Confidence computes the probability that a rule will hold. It is the ratio of the number of transactions containing this combination and the number of transactions involving a single item (antecedent only). It is particularly useful when determining viable product placements. You can also see this application in an online setting. For instance, when you check out on Amazon, you will notice a note that says, “Customers also bought this item…”. It reminds the buyer, if not, conditions them to purchase the suggested item based on its high confidence.

Finally, the lift gauges how strong your rule is in producing a particular effect which is the next purchase product. A lift value higher than 1 indicates a strong rule.

Let’s use the examples below.

Rule 1

peanut butter = strawberry jam

Support – 5% Confidence – 33.% Lift – 1.01

Rule 2

toothpaste = dental floss

Support – 14% Confidence – 44% Lift – 6.77

Rule 3

cooking oil = garlic

Support – .91% Confidence – 33.09% Lift – 8.34

Comparing the different elements from all three rules, Rule 2 appears to be strong, given the high support, confidence, and lift. Simply put, it is worth promoting these products, whether individually or as a promotional bundle.

Rule 1 has high support and confidence but is low in lift. It means these items are selling, but it does not necessarily indicate that the customer will buy strawberry jam after peanut butter.

When support is low, as in the case of Rule number 3, but shows high confidence and lift, it reflects weakness in the sales numbers, not in the association or rule itself. While customers who buy cooking oil will almost always include garlic, the low support means that there aren’t too many people buying them anyway. Low supplies may cause it, or it’s time for the retailer to promote them heavily to warrant attention from shoppers.

Where to use market basket analysis data

1. Product placement

Knowing which products are likely to be bought together, you can position them beside each other, whether in a physical or online store.

2. Marketing

When you find that two products are supposed to go together, but the results of their market basket analysis say otherwise, there must be a better explanation. And it could be that the other product is treated as an alternative to the other. As such, you can promote it when you run out of the primary product.

3. Discover more rules

With the help of your computing resources, you can compare data over any period and select attributes to look into using the reports generated based on a market basket analysis.

4. Cross-selling

Since market basket analysis lets you in on client behavior, you can cross-sell products when the buyer checks out. It’s not all the time that a customer would get donuts after they buy coffee. Hence you can suggest it, and sometimes they need that nudge to realize that they have to buy them.

5. Inventory management

When you have identified the high support items, you have to have enough stock. While marketers may strategize by keeping supplies low of certain products to increase demand and ultimately its price, there are times you need to do the opposite. With associated products, you get to increase your sales on your “if” items and your “then” items.

6. Content recommendation engines

We mentioned earlier how Amazon does it when you check out your cart by feeding you suggestions. Or when you just finished watching on Netflix, you see all these suggested shows. These companies use a software system to learn about preferences using computer learning and integrated statistical modeling to generate predictive content.


Retailers can use computing resources to provide them with comprehensive market basket analysis. It is a great technique that allows you to direct your marketing efforts more precisely.  If you do it right, you hit two, three, even four birds with one stone, which can help achieve your sales targets.