‹ back to all posts
Breaking Down Promotional Email, Part 2

Breaking Down Promotional Email, Part 2

Segment Based "Personalization"

Good afternoon, reader - the wait is over!  Week two of Jetlore's promotional email breakdown is finally here!  

Our last post discussed the "one-size-fits-all" email, where retailers create broad, discount heavy promotions that they send to all of their customers.  (Read last week's post here.)  

Today, we're taking a look at another very common practice in promotional email - segment-based recommendations.  The examples used are from my (Chris McCarthy, Content Marketing Manager) personal email account; this week, the emails are all from one of the first e-commerce sites ever, a global marketplace with millions of subscribers, products, and resellers. 

Smaller retailers often try to imitate this company's style of recommendations because of their success and longevity - but are their product recommendations actually successful or even good representations of a positive user experience?  Or is their success due to the fact that they've utilized SEO and a comprehensive re-seller platform to become the first destination for any shopping need? Take a look at the examples below, and let us know what you think.  

A Quick Re-cap on Segment-based Recommendations:

Segment-based recommendations rely on very recent user action to place customers into a segment, sometimes known as a "buyer persona" or "bucket."  Marketing teams define segment rules based on actions or clicks, and when a user fulfills that rule, they're placed within a specific group of customers that took the same action.  The amount of segment rules and how granular marketing teams get with segments dictates the amount of work that follows.

Often, retailers build creative content for each of their segments similar to "one-size-fits-all" emails - just several more of them at a time.  Because of that, they'll create entire "one-size-fits-all" emails for each segment, or emails with a static hero image at the top and blocks of content at the bottom specific to each segment.  Merchandising for all of those various segments is still a great deal of work;  to make these emails more scalable, this retailer also utilizes a form of automated content generation called Collaborative Filtering. 

The idea behind collaborative filtering is that once you're placed within in a segment, the most relevant content is what the majority of others in your segment showed interest in (i.e. similar users also bought).

Example 1:

The first example we'll provide is this company's use of static segmented content.  


Having purchased fitness gear on this site in the past, this recommendation falls within the correct context, but the copy is generic, the image is static, and the merchandised content is the same for me as everyone else in the "Exercise and Fitness" segment.  They couldn't even populate my name with a personalization token!  Ultimately, this is just one step above a "one-size-fits-all" experience by attempting to activate intent through the "deal of the day" discounts.

Example 2:


As we're sure you know from following Jetlore's social feeds, part of our unique culture revolves around our super cool 'work in slippers' mantra.  In an effort to finally be chosen as Jetlore's #SlippersOfTheDay, I purchased a sweet pair of brown men's animal claw slippers.  

This brings me to my first point; three of the items displayed are completely identical to the slippers that I purchased, and another three are exactly the same in a different color.  This retailer must be confusing me with my boss, who literally does own like 30+ pairs of similar slippers. However, with the exception of Girl Scout Cookies (I'll buy an infinite number of thin mints), suggesting I buy the same item I just purchased doesn't make sense.

And although I purchased men's slippers, the email content suggests that I'd be interested in women's and children's products, for which I have no need.  So again, this email is similar to a "one-size-fits-all" experience, where the retailer has found an item type I'm interested in and throws a bunch of related items my way hoping at least one works.

To be fair, as a standalone example, it's not terrible - even though they're no longer relevant, I can understand why I'm seeing these products and that there's a contextual correlation to something I bought. But after my behavior demonstrated I wasn't interested in those product and I wasn't engaging, obviously they stopped recommending those items....

Ohhhh, wait a second - that's right, they actually sent me the exact same email two days later.

Example 3:


While the featured products are ordered differently and the slippers at the top are new, these are all just the updated selections that customers in my segment have chosen (again, my boss and I are clearly in the same segment - I think she has those pig slippers).  

Collaborative filtering technologies can only utilize recent information, so my last purchase is all this retailer knows about me.  Because of that, I'm limited to looking at items that people with that single shared action purchased or clicked on - so until I display another action, I'm going to see more furry slippers (Stephanie, stop buying slippers!!).  Then, that new action will follow me around in exactly the same way, and the process will repeat until I unsubscribe.

The biggest problem with this method is that it is reactionary - meaning they can only suggest things based on what I used to want, and can't suggest items that I will want next.  And really, is it actually "Personalization," if you're stuck seeing the same collection of maybe-relevant goods as thousands of other people?  

Stay tuned - the conversation continues next week....

Breaking Down Promotional Email, Part 3: Going Beyond Personalization to Predictions