How to A/B Test Different Titles for Google Shopping

One of many things Feedonomics enables you to do is easily write rules to parse, include, re-arrange meta data in titles. You can even then see the results of the optimizations through large performance gains.

Now Feedonomics lets you A/B test different titles for engines like Google Shopping, Bing Shopping, and many more!

The process to A/B test different product titles is as follows:

1) First choose a sub-selection of products

This sub-selection can even be all products if desired.

An example would be choosing products where the brand is equal to Black & Decker:

[brand] equal ‘Black & Decker’

2) Write the split test
The unique_id() function will give us a hex string starting with 1-9 or A-F, or 16 possibilities. So if for instance we wanted to only try putting the brand name first for 1/16 or 6.25% of the Black & Decker product titles, we could do something like this:

Rules for new title

if: unique_id([id]) begins_with ‘1’

then: [title] = [brand] [product_name]

Rules for old title

if: unique_id([id]) not_begins_with ‘1’

then: [title] = [product_name] [brand]

3) Set a custom label so that you can judge performance in the Dimensions tab
if: unique_id([id]) begins_with ‘1’

then: [custom_label_3] = ‘brand first’

if: unique_id([id]) not_begins_with ‘1’
then: [custom_label_3] = ‘title first’

4) Analyze the lift

Remember to normalize for your percentage split and use whatever methodology you want to get statistically meaningful data.

5) Profit!