Ultimate product cross promotion module

We all know that cross promoting products is one of the win-win ways to help store owners to increase the total per sale as well as helping customers to quickly locate the items they need (and avoid the cases when they forget something and later have to place another order and pay the extra shipping fee).
One example is from my personal experience, just a few weeks ago I purchased my Ipod Touch, I was so excited by that and totally forgot about purchasing screen protection and cover for it. All those who use Iphone or Ipod Touch know too well how quickly these items can get dirty and covered in scratches without proper protection. I later had to go out to a near by store to buy them.

Who lost in this case? The online store missed the chance to sell more, and I wasted my time.

Given then importance of cross-selling, it’s surprising that the available cross-sell modules for zencart are limited and in 99% the cases done manually by store owners (cross selling the items they think related). There are several problems here, however:

  1. All this is manually inserted by store owners, which doesn’t really reflect exactly what the customers have in mind
  2. These cross sells are not individually oriented, customer A will see the exact items suggested to customer B. In reality, each customer has different reference and we all know that.
  3. For stores with thousands of products, manually cross selling them can be a mission-impossible task (well, not quite, but it does takes much of your precious time)

For all the reasons mentioned, we decided to work on a new cross sell system which we call our “ultimate product cross promotion module“. The goal is to have the system that automatically takes in various elements and then suggest the best items that matched the individual user’s needs

Our development process is broken in 3 phases:

Phase 1

Introducing the relevancy/relatedness scoring system: our old cross sell system allows you to manually insert related products, as well as giving them a manual “sort order” so that you can manually put one at higher position. With the new scoring system you will still have the ability to manually insert related products as well as manually set the position (since most of the time, the machine’s calculation cannot beat a human’s brain). However:
  1. The module now can automatically compare each product against others based on different elements such as: keywords it found in product name, description, meta description as well as other things such as tags to calculate a relevancy score between each pair of products.
  2. It will allow you to give a weight factor to each element (for example, title is most important for you so you give it a weight of 3 while tags only has weight of 1)….You can of course disregard certain element by giving it a weight of 0.
The good news is that phase 1 is basically finished and it will be free for all. We will send emails to our newsletter subscribers soon so that you can test out this new system, stay tuned.

Phase 2

Go far and beyond: we will not stop at static items like title, description, we will use dynamic elements such as the usually purchased together products, viewed together products (in the same session) as well as other elements.
This is the phase where the product suggestion will go to individual level, and each customer may see a different list of products suggested to them.

The final product of phase 2 will be an open-source commercial product, it will be reasonably priced as our products, well supported and will be worth it for the stores that need them (if your store has less than 100 products for example, you can utilize the free version with all the features listed in phase 1 and probably have no need for a more advanced version)

Phase 3

More tweaks, more analysis: this is the phase where we use all your feedbacks to tweak our scoring system as well and completing a detailed analysis in the backend. In the end, what is the use of this system if you don’t know exactly how good does it perform? Here you will know the % of customers click on the suggested products, the % customers that do add the suggested products to cart and then go through the checkout etc…. Various types of analysis data will be provided to you so that you can better understand your customers, tweak your cross sell listing, or make changes in your current stock to get more items that they want.

The final product of phase 3 (and after) will be an upgrade to phase 2, so if you own it, you are free to upgrade.

A few notes:

  • Performance: there is nothing to worry about performance, the score calculation process is not done per customer visit, it is set to be run automatically (using cron) at the best time on your store(late night, for example)
  • Multiple product promotion list: we have been talking about cross selling only, but how about simply suggesting products to customer individually? For example, Amazon.com can suggest new items I may be interested in each time I visit the store based on what I bought, reviewed, added to wishlists etc… That is something we consider adding either in phase 3 or 4.

One Comment

  1. Isaac says:

    This is very exciting and we are definitely interested in helping test out this mod and provide input as needed. Thanks for all the great contribution to the Zen Cart community.


Leave a Reply