Decoding Buying Behavior – Product Affinity Analysis Anirban Majumder March 11, 2021

Decoding Buying Behavior – Product Affinity Analysis

Every retailer is actively trying to increase their revenue generated per customer or what is frequently referred to as RevPAR. The first and apparently the most obvious method employed in traditional retail is to deploy more advertising to entice the customers. But how do you choose which product must you deploy your marketing spend on? Ever noticed how Amazon magically offers you recommendations for items you are possibly looking for?

Allow us to introduce you to product affinity analysis or market basket analysis, a technique pioneered by Amazon that has moved into prominence now more than ever.

Affinity analysis, in a nutshell, is a data mining technique used to find co-occurrent relationships. A simple example of this would be someone going out to buy shampoo might also buy some conditioner. This simple example illustrates the relationship between the two products. These examples have become more widespread than ever with us getting suggestions about other products that we might consider buying along with the primary product.

What is the need for product affinity analysis in retail?

Now, more than ever, retailers are relying on data analytics to help increase their sales in stores, support their manufacturing processes or other such process improvements depending on the market segment and the sector that they belong to.

One of the solutions that retailers are increasingly looking at in data analytics is product affinity analysis, our topic of discussion in this blog. Even though it has been around for a considerable amount of time, it has not been effectively used by many retailers. By investing in the correct solution, a retailer can gain the right insights about its customers. These can help the retailer to decide the merchandise to be offered, besides helping in designing loyalty programs, sales promotions, discount plans and up-selling and cross-selling tactics.

For instance, why must you run a sales promotion for two different products when a promotion on just one will drive sales in the other due to the high probability of these two products being purchased together? There is a reason why, when you go into stores, you see products on discount prominently displayed next to items that they have a co-occurrent relationship with. As more retailers decide to take an omni-channel approach, they need to understand not only the features that are being offered by the software solutions but also analyze how they can increase sales by leveraging advanced analytics methods such as affinity analysis.

Everyone remembers the famous case of Target, a US retailer, who unknowingly sent pregnancy and baby product information to a woman before she had a chance to inform her family of the same. The retailer was able to leverage the data it had with regards to the purchases made to come to this outcome. This example underscores just how powerful the data that retailers have access to is. It opens up opportunities to increase sales via better isle placements for products, understanding customer behavior, develop cross-promotional programs and all of this from a central dashboard that helps you actively pull and push levers for products to maximize their performance.

What should a retailer look for in a software for affinity analysis?

The relationship between products might look simple and straightforward, thanks to the example mentioned earlier vis-a-vis the shampoo and the conditioner, but its implementation is fairly more complicated than it would seem. Let’s assume that there’s a simple chain of grocery stores with 50 outlets in the network. These 50 outlets would easily process more than 10,000 transactions per day and if they have an online presence, that would add another 60,000 transactions per month. The solution that you deploy for affinity analysis in this scenario has to be robust enough to handle this kind of data inflow on a daily basis. I agree that this example is an oversimplification of the data problem, but I want to drive the point home. Just imagine the kind of data a larger grocery retailer with 100 stores would have to handle every day. We are talking about a mindboggling data set.

Beyond the ability to handle enormous sets of data, what retailers should also look for is the ability to handle the data analysis in an expeditious manner. The greatest pitfall most retailers fail to consider is that affinity analysis holds true for not more than one year. The time frame could be lesser in the case of products with seasonal constraints. For example, the likelihood of face masks getting sold along with hand sanitizers during our current pandemic. The product you choose should be able to handle active analysis of the data to provide you actionable intelligence.

Another feature that you must keep an eye on is the ability to wrangle data at your end as a user. It is fair to assume that the user would want to break the data down to view it from multiple perspectives. These slicing views provide you the power to analyze specific data that can help you isolate the high sales-generating product or to find the problem factor. This kind of a visibility into your tiniest dataset is nothing short of having a God view into your business operations. You are the expert in your business and the solution you pick should enable you to explore data as you deem fit and not hinder your ability to review your retail performance. At Prescience, we have been able to integrate all the above and more for our clients. We are not only able to integrate slicing views to help you understand the data generated better, but we are also able to handle the large sets of data better. This is due to the optimized calculation matrix we have utilized in the form of sparse matrix (if you have questions about sparse matrix, we are more than happy to explain it). Our approach works even for an ecommerce marketplace, something that a lot of retailers have been working on lately to go directly to the consumer. The ability to break the data down to a product promo level, from a macro view to a micro view, will help you better utilize your time and ensure that you can continue to grow your business.

If you are looking for a starting point for your business, take advantage of our personalized FREE consultation workshop 

Subscribe for more Gen AI updates: Stay ahead with the latest on Gen AI by joining our mailing list.