Last Updated: 14/11/2018
- 1. Email engagement bands
- 2. What interactions are counted in engagement score calculations
- 3. How is the Engagement Score calculated
- 4. Contacts with a low number of interaction events
- 5. How is the average engagement score calculated
Assigning contacts email engagement scores, helps you to identify which contacts are engaged and therefore:
- send emails only to only engaged contacts and not batch and blast to everybody, to avoid any deliverability issues.
- use the email engagement bands to accurately segment on the basis of contact specific interactions, so you can optimise the customer’s experience.
Naturally, the first step is to accurately identify the level of engagement for each contact.
1. Email Engagement Bands
Each contact has an email engagement score explained in: How is the engagement score calculated. Based on the engagement score, each contact is placed in an email engagement band.
The email engagement bands are calculated over a 4 week period so you track contacts who have subscribed recently but have also had enough time/opportunity in 4 weeks to engage with a higher volume of email activity.
- Null: means that there is no email engagement data for a contact, so an engagement score cannot be calculated. This is not the same as 'no engagement'; it just means there is no engagement data for the contact in the platform.
- New: is for contacts that have subscribed within the last 4 weeks and have not interacted with (opened or clicked) an email. Once those people start interacting they are moved to their respective interaction bucket. E.g. if the contacts have not interacted with any emails even after 4 weeks they are moved over to the unengaged/ no engagement band.
- No engagement: is for contacts with engagement score = 0. For new contacts: if it has been 4 weeks since they have received emails and have not engaged with any email, they are classified under no engagements. For old contacts: if It has been 180 days since the contact last engaged with an email they are classified under no engagement. The unengaged metric is particularly useful sending and scheduling broadcast campaigns, so you can exclude unengaged contacts in the Recipients tab.
- Low: Engagement Score < (0.5 * average score). If a contact has an engagement score that is less than half of the average engagement score of all your contacts, Ometria puts that contact in the low engagement band. Unengaged contacts that start engaging with emails are automatically transferred to the low engagement band. The average score is account-specific. So the average engagement score of your company’s account would be different than other retailers average scores.
- Medium: 0.5 * average score < Score < 1.5 * average score
- High: Score > (1.5 * average score)
Note: Dynamic segments update every 8 hours so and engagement changes might not be immediately reflective. A good practice would be to check it the next day.
2. What interactions are counted in engagement score calculations
Before April 30, 2019, we only counted email opens/clicks for a contact if the contact opened/clicked a campaign email in 3 days after the email was sent.
So if a contact interacted with an email on the 4th day after receiving the email, the interaction event would be updated in their contact profile on the platform, but would not be used to update their engagement score.
This interaction-within-3-days-of-sending limit was in place for performance reasons.
After April 30, 2019, however, the limit is extended to 7 days. So now, even if a contact interacts with a campaign email on the 7th day after receiving it, the interaction will not only be updated in their contact profile, it will also update the contact's engagement score.
- This means that the number of people in engagement bands before April 30 will not change, because they have been placed there by the previous logic. And the new update has not affected historical data.
- Moving forward, for campaigns sent after April 30, clicks and opens in 7 days after receiving the email, WILL be included in updating contact engagement scores and hence update their engagement band.
- There will always be people with 'no engagement', but with clicks/opens events. Because these people could have interacted with the email after 7 days of being sent the email and the interactions would not have affected the engagement score. (3 days before April 30)
- There will also, always be some people with a positive engagement score, but with no click/opens, if they have had less than 6 opportunities to engage. This is because they are given account-average protection. See Contacts with a low number of interaction events.
3. How is the engagement score calculated
All engagements are based on interactions or events that happen inside Ometria after acquiring a contact.
An engagement score is calculated by looking at deliveries, opens and clicks in a set time window for each individual contact. This window is the last 180 days, expressed as window size (days).
Note: Transactional email interactions are not included as part of email engagement calculations.
- For each email sent to a contact, we calculate a basic interaction score by weighing the number of opens and clicks and multiplying that by an age decay factor calculated for that interaction.
- Clicks have a higher weighting than opens as we consider a click to indicate a higher level of intent by the email recipient.
- The age decay factor takes into account not only the number of email events but also the recency of these events. It is a linear function based on how many days ago the interaction occurred. We apply this age weight to a score so that more recent contact events have a greater weighting than older events.
- For each contact, we then calculate a raw engagement score as a sum of all their interaction scores, divided by the sum of the interaction weights and sum of all age decay factors. This raw engagement score is a weighted average of email interaction and recency of the contact.
- Important - the advantage of this enhanced calculation is the older the email interaction data, the lesser its value compared to more recent email interactions. Additionally, as the sum of scores is divided by the sum of weights and decay factors it prevents old data from decreasing in value as long as there are no new delivery events.
4. Contacts with a low number of interaction events
- All of the contacts have a set number of minimum required events. If they are below that limit they get an account average protection.
- The account-average-protection formula prevents people with less engagement that have a raw score of nearly zero, to have non-zero values that can classify them in an engagement band.
- If a contact has not received the minimum number of events in the time window, an account-specific average is proportionally applied to ensure contacts who have not had the opportunity to engage are not defined as unengaged. For example, if a contact has not been targeted in any recent campaigns this will not disproportionately impact their score.
5. How is the account-average-engagement score calculated
Account-average-engagement score is just the sum of the raw engagement scores for all contacts divided by the number of contacts that have been assigned raw engagement scores in your Ometria account.