Welcome and nurture sequences

Engagement Scoring Model for Email Marketing Teams

An engagement scoring model turns opens, clicks, replies and bookings into one number per subscriber, so you know who to mail, when, and with what message.

Updated 9 min readBy Trevor DavisSegmentation & Personalization

An engagement scoring model for email assigns each subscriber a numeric score based on their behaviour, things like opens, clicks, purchases, and site visits, so you can prioritise who gets your best campaigns, who needs re-engagement, and who should be suppressed before they drag down deliverability. It's the most reliable way I know to turn a flat list into a ranked audience you can act on.

The question it answers is simple, and most dashboards can't: who on this list wants to hear from me right now? For example, say a consultant rolls opens, clicks, page visits and enquiries into one number and ranks the whole list by it. A small group at the top is likely doing most of the engaging. And a group that looked promising on paper, people who filled in an enquiry form, might score near zero because nobody ever followed up with them. The score ranks the list, and it also shows you where the leak is.

That's what an engagement score does that source tags and demographic buckets can't. Those approaches tell you where a subscriber came from, not where they are headed. An engagement score tells you something more valuable: how much this person wants to hear from you right now. That signal drives better timing decisions, smarter suppression, and campaigns that convert because they reach the right people at the right moment.

This guide walks through how I build an engagement scoring model from scratch, the signals to use, how to weight them, and how to put the score to work in your everyday sends.

Key takeaways

  • An engagement score adds up a subscriber's recent opens, clicks, replies, page visits and purchases into one number that shows how interested they are right now.
  • Weight clicks above opens, and replies, enquiries and purchases above clicks, because Apple Mail Privacy Protection makes opens unreliable.
  • Build in decay so the score drops when someone goes quiet.
  • Group scores into four bands and give each band a different kind of email, from offers at the top to suppression at the bottom.
  • A score only earns its keep when it changes what you send next week.

Why does engagement scoring beat static segmentation?

Static segments, the "customers" and "leads" and "subscribers from the summer promotion" buckets, capture a fact about a subscriber that may have been true six months ago. They do not tell you whether that subscriber is engaged today.

Engagement scoring is dynamic. Every time a subscriber opens an email, clicks a link, visits your work-with-me page, or makes a purchase, their score updates. When they go quiet for 60 days, the score decays. The model reflects the current state of the relationship, not its origin.

Here's where I'll take a position that annoys some people. If your score doesn't change what you send next week, it's decoration. It's easy to spend a quarter perfecting weights and never once suppress a dead segment or fast-track a hot one. The score only earns its keep when it drives an action, and ideally an action that runs without you.

For lists of any size, engagement scoring eliminates the guesswork behind "who should receive this campaign." The score answers that question with data every time.


Which signals go into an engagement score?

A well-built engagement scoring model draws from four signal categories. You do not need all four on day one, but each category you add sharpens the score.

Email engagement signals are the foundation: opens, clicks, replies, and forwards. Clicks are more valuable than opens. A click represents an active decision, while an open (especially with Apple Mail Privacy Protection in effect) can be a proxy event rather than a true engagement. Weight clicks at two to three times the value of an open.

On-site behavioural signals layer in intent data that your email platform cannot capture alone: page views, program or service page visits, booking page visits, and time on site. A subscriber who visited your work-with-me page three times this week is showing buying intent that no email-only model will surface. Klaviyo, ActiveCampaign and similar platforms can pick this up through their site tracking.

Purchase and transaction signals create a fundamentally different band of subscriber. A client who has bought twice isn't the same as a subscriber who has opened twice. Their commitment is far higher. Weight purchase events heavily, and separate first-time buyers, repeat clients, and lapsed clients.

Recency signals govern score decay. A subscriber who was highly engaged six months ago but has not opened anything since is not a high-value subscriber today. Build decay into your model. I'd start by reducing the score by 10 to 20 percent for each 30-day inactive window, so the score reflects current engagement rather than historical engagement.

In my experience, combining recency with activity gives you groups that behave more predictably than either signal on its own, and the same goes for a score built from several signals.


How do you build the scoring matrix?

A scoring matrix translates each signal into a point value. Here's the starting framework I use, to adapt to your list:

Opening an email in the last 30 days: 5 points. Opening in the last 31 to 60 days: 3 points. Opening in the last 61 to 90 days: 1 point. No opens in 90 or more days: 0 points, with active score decay.

Clicking a link in the last 30 days: 10 points. Clicking in the last 31 to 60 days: 6 points. Clicking in the last 61 to 90 days: 2 points.

Visiting a program, service or booking page: 15 points per visit, capped at 30 points per 30-day window. Starting a booking or checkout: 20 points. Completing a purchase: 50 points for a first purchase, 35 points for repeat purchases (the extra signal is lower once the client relationship is established).

Submitting a form or downloading a resource: 25 points. Replying to an email directly: 30 points (a high-intent signal that most platforms track automatically).

Score decay: subtract 10 percent of the total score at each 30-day inactive milestone, applied to the rolling score rather than any individual event.

These weights are a starting point. Your offer, buying cycle, and send frequency will need adjustments, but the relative hierarchy (purchases outweigh clicks, clicks outweigh opens, recency matters throughout) holds across most email programs.


How do score bands turn into what you send?

Once subscribers have scores, the next step is grouping them into bands and mapping emails to each band. Four bands work for most lists:

Band 1, Highly Engaged (top 20 to 25 percent of scores). These subscribers are your most receptive audience. Route your most direct, conversion-oriented emails here: offers, program launches, enrolment deadlines. Do not dilute this band with re-engagement content. They are already engaged.

Band 2, Moderately Engaged (middle 40 to 50 percent). These subscribers are active but not enthusiastic. Prioritise educational and value-dense content that strengthens the relationship before the ask. This is where well-timed nurture sequences earn future conversions. For more on matching content to engagement state, see our guide on list segmentation and tailored messaging.

Band 3, Low Engagement (bottom 20 to 25 percent). Send frequency should drop sharply here. These subscribers need a reason to re-engage, whether that is a compelling free resource, a meaningful offer, or a direct "should we keep sending to you?" message. Run a structured re-engagement sequence before making a suppression decision. Our newsletter retention and churn reduction resource covers this sequence in detail.

Band 4, Suppressed (scores at or near zero after decay). Exclude from all standard sends. Mailing large inactive audiences pushes spam complaints up, and Google's sender guidelines ask you to keep those under 0.3%. Suppression protects your deliverability for the people who buy.


How do you run the model inside your email platform?

An engagement score that lives in a spreadsheet is only a report. For the model to affect campaigns, scores must live inside your email platform as a custom field or tag that updates automatically as new events arrive.

Klaviyo supports this natively through calculated properties and flow triggers built on engagement events. HubSpot allows contact scoring through its lead scoring module, which can be configured for email engagement signals alongside CRM data. ActiveCampaign has contact scoring built in, and Mailchimp's contact ratings give you a rough built-in version if you're not ready to build your own.

The key automation to build on top of the score: a band migration trigger. When a Band 3 subscriber clicks a link or makes a purchase, they should automatically migrate to Band 1 or 2 without any manual intervention. And when a Band 2 subscriber goes 60 days without opening, they should drop to Band 3 and enter the re-engagement workflow.

Review the spread across bands monthly. As a starting target, I'd want roughly 60 to 70 percent of subscribers in Bands 1 and 2. If Bands 3 and 4 together grow past about a third of the list, the problem is usually upstream, in where people come from, how they're welcomed, or how relevant the content is, and score adjustments alone won't fix it.


What are the common scoring mistakes?

Over-weighting opens. With Apple Mail Privacy Protection inflating open rates across iOS devices, opens alone are an unreliable foundation for a scoring model. Clicks, purchases, and site behaviour must carry more weight.

Ignoring score decay. A subscriber who earned 200 points last year through heavy engagement but has not opened in five months is not a high-value subscriber today. Without decay, your model rewards historical engagement rather than current intent.

Building the model once and never revisiting it. Your email changes as send frequency changes, offers expand, and audiences shift. Review your scoring weights quarterly, especially when the spread across bands drifts away from your starting target.

Treating all clicks equally. A click on an unsubscribe link isn't the same signal as a click on a booking link. Set your model to tell link types apart where possible, or at minimum, weight offer and booking-page clicks above general link clicks.


FAQ

What is an engagement scoring model for email? An engagement scoring model is a system that assigns each email subscriber a numeric score based on their behaviour: how recently they have opened, clicked, purchased, or visited your site. The score represents their current level of interest in hearing from you and drives decisions about which campaigns they receive, how frequently they are mailed, and when they are suppressed.

How often should engagement scores update? Scores should update in real time as new events arrive, or at minimum, nightly. Batch updates that run weekly or monthly create lag between a subscriber's behaviour and the campaign targeting that behaviour. Most modern email platforms support real-time or daily score recalculation.

Can a small email list benefit from engagement scoring? Yes. The logic that makes scoring valuable, matching message priority to subscriber interest level, applies at any list size. On a 500-person list, the automation infrastructure is simpler, but the strategic benefit of knowing your top 20 percent versus your bottom 20 percent is just as high.

How is engagement scoring different from lead scoring? Lead scoring is usually a sales tool that combines fit (who the person is) with behaviour to decide whether someone is ready for a sales conversation. Engagement scoring focuses on email and content engagement to decide what to send. The two can sit side by side, and for a consultant with a discovery-call offer, they often feed each other.

What should I do when a suppressed subscriber re-engages? When someone in your suppressed band opens or clicks, whether from a reactivation campaign or on their own, move them to Band 2, not Band 1. A single event after extended inactivity is a positive signal, but it does not confirm a return to their previous engagement pattern. Let the next 30 to 60 days of behaviour determine whether they graduate to Band 1.


What could email do next for your business?

If you have an offer, an audience and a part of your email you’d like to improve, let’s talk about where to start.

Let’s talk about your email

Prefer to write? Use the contact form.