A quick caveat to start: if you’re leveraging machine learning/automated bid strategies in your marketing, do not do any of this! Your audience segmentation strategy should be solely based on 1. differing ROAS/CPA goals between segments and 2. differing creative messaging. If they have the same creative and conversion goal, consolidate at all times.

Both Facebook/Instagram and Programmatic Display Media Buying have comparable targeting abilities (albeit with different platforms and ad units) so businesses should have a holistic approach to audience segmentation in each.

My approach is to always have every single audience segment remain 1) mutually exclusive and 2) centered around the customer funnel.

————————–

There should be three overarching buckets of customers (with many sub-segments, but let’s start with those three):

  • (Acquisition) Prospecting:
    • Targeting new customers who have never visited the site before and never purchased before.
    • Actively Targets: prospecting audiences – typically either lookalike audiences seeded from a top customer file, or third-party data sets based on various interests/customer attributes.
    • Excludes: All First Party Data – inclusive of previous purchasers and previous site visitors.
  • (Acquisition) Retargeting:
    • Targeting users who have visited the site previously but have still not purchased. This is still a part of the Acquisition user journey.
    • Actively Targets: Previous Site Visitors.
    • Excludes: Previous Purchasers.
  • (Retention) Reengagement:
    • Targeting users who have purchased before.
    • Actively Targets: Previous Purchasers.
    • Excludes: None.

————————–

Diving Deeper into Retargeting Audience Segmentation:

There are technically hundreds of data points that remarketing segments could be segmented by (depending on their availability):

  • Recency of visit
  • Frequency of visit
  • Depth of site visit
  • Device preference
  • Geographic location
  • Gender
  • Age
  • Interests
  • …and so many more.

The overarching goal with audience segmentation is to 1) find the combination of variables that have the highest correlation with likelihood to purchase and 2) provide significant enough audience sets with enough conversion volume to allow for bid and budget allocation/optimization.

Typically in retargeting – though I’ll never go into any business with too stubborn a set of rules – recency of visit and depth of site visit are the best proxies to likelihood of conversion for a previous visitor who hasn’t converted. Device is useful when segmenting for display as creative, viewability, and likelihood of purchase is particularly volatile on mobile, but it’s less necessary in Facebook/Instagram.

In choosing which segments to choose, it really depends on the size of the available remarketing audience and the number of conversion events you have to optimize on. Another major variable is your willingness to trust algorithmic optimization. Theoretically, Facebook, Google, and every other media buying platform should be able to understand how an audience will behave an auto-optimize in real time. Unfortunately we still haven’t quite gotten there yet but these new tools and strategies are always worth testing.

The notions of incrementality and conversion attribution are still areas where the human touch remains necessary (and thank god or perhaps many of us would be out of a job…).

Here’s a sample retargeting audience breakdown by recency:

  • Visited within 1 day
    • + target users who visited the site within one day
  • Visited within 2-7 days back
    • + target users who visited within 7 days
    • – target users who visited within 1 day
  • Visited within 7-30 days back
    • + target users who visited within 30 days
    • – target users who visited within 7 days
  • Visited 30 or more days ago
    • + target total remarketing audience
    • – target users who visited within 30 days

And here’s a sample retargeting audience breakdown by level of site visit:

  • Visited Home Page only (didn’t progress further on-site – 1st layer of depth)
    • + target users who hit the home page
    • – target users who hit any category/product/cart/conversion page
    • – target converters
  • Visited Category/Product Page (2nd layer of depth)
    • + target users who his any category/product page
    • – target users who hit the cart page
    • – target converters
  • Visited Cart Page
    • – target converters

And here is what a combined retargeting strategy looks like:

  • Visited Home Page within 1 day
    • + target user hit home page within 1 day
    • – target user hit cat/prod/cart/conversion page
  • Visited Home Page within 2-7 days
    • + target user hit home page within 7 days
    • – target user hit home page within 1 day
    • – target user hit cat/prod/cart/conversion page
  • Visited Home Page within 7-30 days
    • + target user hit home page within 30 days
    • – target user hit home page within 7 days
    • – target user hit cat/prod/cart/conversion page
  • Visited Home Page within 30/+ days
    • + target user hit home page (any time frame)
    • – target user hit home page within 30 days
    • – target user hit cat/prod/cart/conversion page
  • Visited Category/Product within 1 day
    • + target user hit cat/prod page within 1 day
    • – target user hit cart/conversion page
  • Visited Category/Product within 2-7 days
    • + target user hit cat/prod page within 7 days
    • – target user hit cat/prod page within 1 day
    • – target user hit cart/conversion page
  • Visited Category/Product within 7-30 days
    • + target user hit cat/prod page within 7 days
    • – target user hit cat/prod page within 1 day
    • – target user hit cart/conversion page
  • Visited Category/Product within 30/+ days
    • + target user hit cat/prod page (any time frame)
    • – target user hit cat/prod page within 30 days
    • – target user hit cart/conversion page
  • Visited Cart within 1 day
    • + target user hit cart page within 1 day
    • – target user hit conversion page
  • Visited Cart within 2-7 days
    • + target user hit cart page within 7 days
    • – target user hit cart page within 1 day
    • – target user hit conversion page
  • Visited Cart within 7-30 days
    • + target user hit cart page within 30 days
    • – target user hit cart page within 7 days
    • – target user hit conversion page
  • Visited Cart within 30/+ days
    • + target user hit cart page (any time frame)
    • – target user hit cart page within 30 days
    • – target user hit conversion page

————————–

Diving Deeper into Prospecting Audience Segmentation:

Prospecting is much more straightforward than retention and retargeting – the same principles of mutual exclusivity apply though. The overarching exclusion on all prospecting audiences should always be a company’s first party data (all previous purchasers and visitors) as these users are taken care of in retargeting and re-engagement.

In both Facebook/Instagram and Programmatic Display, audience overlap must always be mitigated as much as possible. The most complete (but time consuming to a degree that likely isn’t recommended) way of doing this is to, similar to retargeting, create a hierarchical series of exclusions. I’ll dive more into prospecting audience testing in my post on Scaling Performance Marketing.

————————–

Diving Deeper into Re-engagement Audience Segmentation:

This approach is fairly similar to retargeting as well, except the page funnel element can be removed. The first important segmentation variable to consider is recency of purchase. Of note – depending on the business there may be some value to adding some additional exclusions based on this recency. If you have a seven day shipping timeframe, there’s little benefit in messaging customers who just purchased and haven’t yet received their order.

To provide some more food for thought – the other element that could be added here is lifetime value. Both LTV and recency of purchase are excellent segmentation variables to test when deciding on re-engagement segmentation strategy.