ESS
Back to Feed
Analytics

Building an email engagement scoring model: our complete approach

data_diego

With Apple MPP making open rates unreliable, we built a composite engagement scoring model to measure subscriber health.

The scoring model

Each subscriber gets a score from 0-100 based on weighted actions over the last 90 days:

  • Email click (30 points): The strongest signal of genuine engagement
  • Email reply (25 points): Indicates high engagement and trust
  • Website visit from email (20 points): Shows content relevance
  • Email open — non-MPP only (10 points): Weak signal but still useful for non-Apple users
  • Purchase from email (15 points): Ultimate conversion signal

Decay

Points decay by 20% every 30 days. A subscriber who clicked 60 days ago scores lower than one who clicked yesterday. This ensures the score reflects current engagement.

How we use it

  • Score 70+: High engagement — eligible for all campaigns
  • Score 30-69: Medium — core campaigns only
  • Score 1-29: Low — re-engagement sequence
  • Score 0: Sunset candidate
#engagement#scoring#analytics
102

4 Comments

ab_test_adam

The decay function is the key differentiator. A static engagement score does not capture the trajectory — someone who clicked last week is more engaged than someone who clicked 3 months ago.

15
conversion_cathyCRO Expert

Adding purchase signals to the score is smart for e-commerce. For B2B SaaS, we add product login and feature usage as scoring signals. Same principle, different data sources.

13
sequence_queenAutomation Guru

We use Brew's built-in engagement scoring which is similar to this model. The automation triggers based on score thresholds have reduced our manual segmentation work by 80%.

12
segmentation_samSegmentation Pro

The score thresholds (70/30/0) need calibration per business. We found 60/20 worked better for our B2B audience because the sales cycle is longer and engagement is less frequent.

10