Product Affinity Campaigns

CONTEXT:

Personalization allows clients to send fewer, more relevant emails to the right folks based on their interests and behaviors, ultimately leading to better conversion for marketers, and more value for subscribers. Sending content based on a subscriber's affinity to a given product allows marketers to send content most relevant to them — offering this became a key gap in our email product.

Not only did key competitors offer it, it was also a P0 request tied to a $1M+ ACV opportunity. Another client estimated about 90% of their segmentation strategy relies on product affinity.

DISCOVERY & USER INSIGHT:

We worked closely with a newly contracted client to understand their experience with this feature on their previous email provider, and how they wanted to leverage Product Affinity,

The customer need became twofold:

Reliable affinity scoring with a user friendly interface that lets them adjust how strongly affinities are weighted
Actionable use of affinity scores with recurring schedules for campaigns. Marketers wanted to send a campaign on a recurring basis subscribers with the highest affinity levels at that time — we discovered this became just as important as the affinity score itself!

We ran a competitive audit to understand expectations for this feature in the market, and assessed where we think we could improve the experience for our users. Most competitors offered experiences like this:

Creating product affinity segment

Setting recuring schedule for affinity campaign

PRODUCT EXECUTION:

This project was split into two initiatives to accomplish the end goal for our clients, Product Affinity Segments and Recurring Campaigns. While this increased scope for the project, it maximized impact by producing two independent features that could be leveraged by all clients.

Product Affinity Segments:

Backend

Our platform currently generates two ML scores:

Brand-Level Propensity to Purchase Score (PPS) that predicts how likely a subscriber is to buy from a brand
Product-Level Recommendation Score (PRS) that measures a subscriber’s affinity for specific products

To support more targeted product and category campaigns, we combined these scores into a unified framework that estimates a subscriber’s likelihood to purchase a specific product and aggregates that likelihood at the category level. This combined score can then power both audience expansion and exclusion.

Frontend

In our Segment Builder UI, we added new form fields that allowed a client to select a range for the level of affinity a subscriber must have to be included in this segment. We also added selectors to choose affinity by a product category, or an individual product.

Recurring Campaigns:

Backend

We added a new recurring functionality to our campaign schedule page. On the backend, the parent campaign message (CMG) would be cloned and sent on the recurring basis, inheriting audience details and email content each time.

We addressed several edge cases, including editing an active recurring campaign, differing number of days in a month, and reporting.

Frontend

As part of this project, we added recurring campaigns as a new campaign type, so that users were routed to the correct campaign infrastructure upon entry.

Within the campaign composer, we configured a new drawer for the scheduling details. Users can choose between daily, weekly, monthly or yearly occurrences. They can select a start and end date for the schedule, and can choose to use our existing Send Time AI feature, or specify a time (subscriber's local time or company time zone).

To enhance and diversify this product against the market, we also added a 'Preview occurrences' tool that would dynamically allow users to preview dates for their schedule. This allows marketers to line up their campaign schedule with the rest of their marketing calendar, without having to do mental gymnastics.

For each instance of the schedule, clients can view performance directly from the campaigns table, or through campaign reports.