Customers buying 1 product line - what should they buy next?

[rmwillis1973]

Hi,

i have a list of customers who only buy 1 product line (product hierarchy).  i'm trying to work out what they are likely to buy next.

I've extracted a listing of customers who are buying 2 product hierarchies and want to show which lines are best associated with each other.

I've gone through a number of videos and the MB Affinity / MB Rules / MB inspect and can't seem to get what i need.

I'd like to be able to display a priority listing of products:

Customers who bought product 4 also bought product 23, followed by product 7, product 2 etc with an associated score of likelihood.

i can obviously do this in excel, but am working with large dataset.

[JohnJPS · Accepted solution]

Hi @rmwillis1973,

Just a quick clarification: would you consider an association of "4 followed by 23" to be the same as "association group" as "23 followed by 4" or should it be a separate item in the final analysis?

The attached wrokflow assumes they're the same and produces counts accordingly.  Is that close to what you're after?  There are no MB tools at all, just some slicing and aggregating...

[rmwillis1973 · Accepted solution]

Hi,

I used the principles you showed me to make a neat solution.

I was able to find anyone who had purchased 2 or more categories, rank the number of orders containing a category and then select their 1st and 2nd preferences.

Therefore for customers who have only bought 1 product category, i can make a suggestion for their "next" category, based on what people who look like them have purchased.

I was also able to do this by country, so that the "next" category differs by country.  Again, a neat solution thanks to your help.

I've messed around with lots of MB Affinity / Rules including the Sample module.  Whilst it does help visualise the data, it doesn't help me as much when i need to see patterns across countries.

Thanks again!!

Richard