Basket analysis: using purchase patterns to increase sales

Basket analysis is the practice of studying what people buy at the same time. For example, if grocery shoppers frequently buy crackers and dip together, the store can use that information to run combo deals or place those items next to each other.

The same concept is also used for online recommendations. In fact, 35% of Amazon purchases and 75% of Netflix views come from suggestions based on basket analysis.

In this article, we’ll unpack what basket analysis is, why it matters, and how to use it to get more sales and happy customers.

Basket analysis looks at what customers buy together to find patterns. Companies can then use those insights to recommend related products to encourage more sales. This practice is referred to as ‘cross-selling’.

In a physical store, recommendations might come in the form of a combo deal or simply by placing the items near one another. For example, a bike shop might notice customers often buy helmets with new bikes, so they display them side by side or offer a bundle discount.

In e-commerce, this shows up in the ‘frequently bought together’ section, where shoppers are recommended additional products based on what others purchased together.

Basket analysis provides the insights, and personalisation puts them into action by recommending items that fit each customer’s habits. When done well, this can reduce acquisition costs by up to 50%, lift revenue by 5% to 15% and improve marketing ROI by 10 to 30% .

Now that we’ve covered the basics, we can look at the step-by-step process that turns customer transactions into patterns you can use for your future sales.

The process always starts with data collection. In retail, this could be point-of-sale receipts; online, it might be ecommerce carts or subscription records. Each transaction shows which products or services were chosen together.

Having accurate data is important because errors can lead to wrong recommendations and distort the results.

Once the data is collected, algorithms such as Apriori or FP-Growth scan the data to find products or services that frequently occur in the same transaction. These groupings are called ‘frequent itemsets’.

From those itemsets, the system creates rules using an “if this, then that” format. For example, a rule might be: If a customer buys lipstick, they are likely to add the matching lip liner.

Algorithms calculate three key metrics and then apply the strongest rules in recommendations or promotions:

• Support: How common it is for two items to be bought together
• Confidence: How likely someone is to buy item B when they’ve already bought item A
• Lift: Whether people buy the two items together more often than you’d expect by chance

Following these steps turns raw customer data into insights your business can use to increase sales by delivering the right recommendations at the right time.

Once you have your itemsets and rules in place, you’ll want to put them into action. Companies across industries use basket analysis insights to do the following.

Businesses can place frequently purchased items near one another, both in-store and online. For example, supermarkets often put taco shells next to salsa to make the buying decision easier.

Companies can design promotions around products that are often bought together. For example, a gym equipment retailer might notice that customers who buy dumbbells frequently add resistance bands. They could run a promotion that bundles together strength training equipment.

Retailers can ensure that products commonly bought together are always stocked at the same time. For example, a hardware store might notice that customers buying paint often buy brushes, so both need to be replenished at the same time.

Traditionally, cross-selling and upselling relied on sales representatives suggesting add-ons or upgrades. Basket analysis makes this data-driven by suggesting the products customers buy together. For example, if camera buyers often add memory cards, the store can surface that recommendation automatically at checkout.

While the most obvious examples of basket analysis come from supermarkets and online stores, the same technique can be applied across many industries to uncover patterns and increase sales.

Here are some of the most common applications by industry.

Salesforce customers across Australia are using data-driven insights to personalise experiences and boost growth. Here are two examples to inspire you.

Barbeques Galore has been a household name since 1977. However, when the retailer launched its digital storefront, it needed a way to bring the same personal touch customers experienced in-store to the online world. To achieve this, they turned to Commerce Cloud.

Using Einstein Product Recommendations, powered by basket analysis, their website can now suggest items frequently bought together. This means that customers browsing online see tailored product suggestions alongside real-time stock updates, making it easier to find everything they need.

Since implementation, online sales have doubled, and customers now enjoy a shopping experience that feels as helpful as being guided by an in-store expert.

True Alliance manages some of the world’s most recognisable brands in Australia and New Zealand, including The North Face, Speedo, and Coach. Its role is to help these global brands succeed in the local market by running their e-commerce sites, retail stores, and wholesale operations.

As their portfolio grew, True Alliance needed a way to scale quickly and deliver consistent, personalised experiences across every channel. To achieve this, it adopted Salesforce Commerce Cloud, Marketing Cloud, and Service Cloud.

With these platforms, True Alliance built a complete view of shopper behaviour across its brand portfolio. Using Commerce Cloud’s basket analysis, the company can now suggest products that are often purchased together, both online and in-store, helping each brand increase sales.

Since implementation, online sales have increased by more than 200% year on year.

Basket analysis can be done in different ways, from a single analyst coding in Python to large companies using integrated tools like Salesforce.

The right choice for you will depend on your team’s skills and the scale of your company. Here’s how the main approaches to basket analysis compare:

To get the most value from basket analysis, you need to treat it as more than a data-gathering exercise. Here are some best practice tips.

• Use organised and accurate data to avoid skewed results.
• Set thresholds for support, confidence, and lift so you only act on meaningful patterns.
• Tie every rule to a clear business goal, like lifting average order value.
• Test recommendations in pilot campaigns before rolling them out more broadly.
• Feed insights into your CRM or marketing tools so you can use the insights across your revenue team.

• Acting on messy or incomplete data that generates false patterns.
• Applying every rule you find, even if the lift is low.
• Creating bundles or offers that logically don’t make sense for your customers.
• Scaling campaigns without testing whether recommendations are increasing sales.
• Letting insights sit in spreadsheets instead of actioning them.

Basket analysis shows you what customers buy together and turns that into recommendations to increase sales and improve the customer experience. When done correctly, it’s proven to grow your average order value, boost customer retention, and increase your marketing ROI.

Using Salesforce Commerce Cloud, which offers Einstein Product Recommendations, you can put these insights to work quickly and make it easier to give customers what they want.

If you’d like to see Commerce Cloud in action, you can watch the free demo here.