Skip to Content

Retailers are aware of the importance of product placement and the effect it has on customer’s purchasing decisions. Suppliers are fighting over prominent shelf space and retailers spend many resources in an effort to optimize the product placement so it will fit most customers’ needs and increase customer spending. However, the products arrangement at retail stores is static and remains the same for all customers stepping into the store. Imagine how much more could be achieved if you could personalize the product placement according to each customer’s behavior and characteristics…

This marketer utopia has realized with the development of the world-wide-web and the spreading of web-shops. Now not only does the marketer have the option to modify and personalize the product offering in each customer session, he can also follow each customer’s behavior in the web and has a vast amount of accurate actual customer behavior data available at his disposal.  

This data rich environment can be used by the retailers to optimize the product offering in each customer sessions.One of the common uses for this data is market basket analysis in the form of – “users who bought x also bought y”. These recommendations are based on item-to item relationships between pairs of items that are likely to be purchased together. Theoretically, if we know for each item what are the other products with which it is most frequently sold we can recommend them on the site and increase the chances of the customer purchasing an additional product. 

A friend of mine used to work for Amazon UK shared an amusing story: The products lists displayed in the site were automatically generated from a data warehouse and they got a complaint that the “customers who bought x also bought y …” lists for large screen TVs brought up products from the adult film categories, and it was not due to a problem in the data analysis…  

This story brings up some questions about what factors should be considered when selecting the related products lists that are presented in the site.We can see that the product category might be information that should be taken considered. While it might be ok to suggest large screen TVs to customers interested in adult movies, the opposite recommendation from the same association is not always ok.  

The purpose of the related products lists is to encourage the user to maximize his current and future spending in the website. We would like to offer the customer a large selection of products and categories, hoping that it will increase the chance of him finding something to his liking. However, presenting too many different options could confuse the customer, who might have initially entered the site looking for a certain product but finds himself following a chain of links that bounces him from one category to another. We do not want the customer to feel lost in the site or for him to get the feeling that it is hard for him to find what he is looking for. 

Another consideration is the type of product that is offered in each stage of the session. A good example are batteries, this is a product that can be sold along with many other products, but if we got the customer looking for a product in our web shop, we don’t want him to end up buying only a package of batteries. We would like to bring up other offers and suggest the batteries to him only at checkout, after he completed his main purchase. The context and the session flow can also affect our recommendations. Another consideration that might be taken into account is the price of the products that are offered throughout the session. Naturally, we want the customer to buy the more expensive items, but we do not want to present products that are too expensive in a point where the customer would not be interested in spending more, and by that losing the client and giving the impression that our site is expensive.  

There are many other things that might affect our selection of web recommendations. You might have noticed that I haven’t even addressed the issue of personal customer information that can also be used for the product recommendations optimization. 

I’d be happy to hear your ideas about factors that should be considered to make the product lists in the website more affective. Please feel free to share your thoughts in the comments sections.

To report this post you need to login first.

2 Comments

You must be Logged on to comment or reply to a post.

  1. Amir Baruch
    funny that you write about e-shopping. just 5 minutes ago I bought a mini lap top online, and I swear that 15 minutes ago I didnt even think I need one… so what could encourage impulsive purchase??
    I can’t say that the lap top is a good example as I just bought one item and refused to add accessories, but here are some thoughts:

    1. show items that were bought along side the item you are now reviewing, by other customers.

    2. the first option already exist in many sites, but what if you could show only those items that were bought by customers that fit your cluster (age, location…)

    3. I believe you should only suggest items which are as expensive or less expensive than the one you are looking at. dont remember when I bought a second item which was more expensive than my first.

    anyways liked the post.
    Amir

    (0) 
  2. Yaron Livneh
    So the whole concept is based on the assumption that if you’ll push people will buy (even if they didn’t plan to buy¡K)?
    I didn’t think it can work until I have seen the previous comment ƒº
    Clustering seems to be the most intuitive way to push the right products, but the million dollar question is how to cluster?
    Customer characteristics probably cluster not bad (if you can define the most common), but might be not sufficient as there are other environmental factors such season, holidays, campaign and more¡K.
    From my experience, the connection between products usually predicts well shopper interest.
    Any way the idea is very interesting,
    Yaron
    (0) 

Leave a Reply