Email engagement score

An email engagement score is a single weighted number that summarises how actively an individual recipient interacts with your email, combining opens, clicks, replies and recency into one figure. A recipient who replied yesterday scores far higher than one who opened a single email three months ago, even though both technically engaged. The score collapses a stream of behavioural events into one comparable value per person.

The score matters because it lets you act at the level of the individual rather than the list. A list-wide engagement rate tells you the overall health of the audience, but it cannot tell you which specific recipients to prioritise, suppress or re-engage. The score can. It is what lets a sales team chase the warmest leads first, a marketing team sunset the coldest contacts, and a lifecycle programme route each person to the right next message.

Because it weights stronger actions more heavily and decays over time, the score reflects current intent rather than lifetime history. That makes it a practical input to lead prioritisation, where it sits alongside lead conversion rate as a signal of who is ready to move. A clearly understood score is more useful than a precise one, so most teams favour a simple, explainable model over a black box.

The score is a weighted sum of actions with a decay term that pulls it down as a recipient goes quiet. The exact weights are a business choice, but the structure is consistent. Set the weights, apply a decay, and recalculate on a regular cadence so the score stays current rather than frozen at the moment of the last big interaction.

1. 1

   Action weights

   Assign points to each action by how strongly it signals intent. A common starting point is one point per open, three per click and ten per reply, then tune from there once you see real distributions.

2. 2

   Recency decay

   Subtract points as time passes since the last action, or apply a half-life so older actions count for less. Without decay, a recipient who was active a year ago keeps a high score they no longer deserve.

3. 3

   Negative actions

   Subtract points for unsubscribes, spam complaints and hard bounces. A recipient who marked you as spam should score near the floor regardless of how engaged they once were.

4. 4

   Normalisation band

   Map the raw weighted total onto a fixed band, such as 0 to 100, so scores are comparable across recipients and over time. A bounded score is easier to threshold into cold, warm and hot tiers.

A worked example shows the moving parts. Suppose opens are worth 1, clicks 3 and replies 10, with a decay of 2 points per week of inactivity. A recipient who opened four emails, clicked twice and replied once in the last week scores 4 plus 6 plus 10, which is 20, with no decay applied. A second recipient with the same actions but whose last touch was eight weeks ago loses 16 points to decay, landing at 4. The raw behaviour is identical, yet the score correctly ranks the recent recipient far warmer. Mapping both onto a 0 to 100 band then makes them directly comparable to everyone else.

A metric tree decomposes the engagement score into the inputs that build it and connects each input to the team that influences it. The score is already a composite, so the tree makes its construction visible rather than hiding it inside a formula. When a recipient or a cohort scores lower than expected, the tree shows which input is responsible.

The first level splits the score into intensity, recency and quality. Intensity covers how many actions a recipient takes, driven by send frequency and content relevance. Recency covers how recently they acted, driven by cadence and the freshness of triggers. Quality covers how strong those actions are, driven by whether the email earns clicks and replies rather than mere opens.

Reading the tree turns a low score into a specific cause. A recipient high on intensity but low on quality opens everything and acts on nothing, which is a content and offer problem. A recipient high on quality but low on recency was strongly engaged and has since gone quiet, which is a re-engagement problem. Each pattern points to a different owner and a different intervention.

Because every team sets its own weights and decay, absolute scores are not comparable between organisations. What is comparable is how you tier the score and what each tier means for action. The bands below show a typical 0 to 100 scheme and how teams usually treat each one.

The useful benchmark is the shape of your distribution, not any single number. A healthy list has a meaningful hot and warm core with a manageable cold tail. If the cold band keeps growing while the hot band shrinks, the list is decaying faster than you are replenishing it, and no weighting tweak fixes that underlying trend.

Improving the average score across your list is less about gaming the formula and more about earning stronger, more recent actions from real recipients. The score is a mirror, so the work is in the inputs it reflects rather than the weights themselves.

The metric tree approach starts by finding which input is holding the score down across a cohort. If quality is the constraint, the content and offer need work. If recency is the constraint, the cadence and triggers need work. Fixing the binding input lifts the score for everyone in that cohort at once.

KPI Tree lets you attach RACI ownership to each branch of the score, so the team accountable for content quality is distinct from the team accountable for cadence, and both are visible. When the average score for a segment drops, the platform pushes the change to the accountable owner of the input that drove it, rather than leaving a number to sit unread on a dashboard. The verified impact loop then confirms whether a change to weights, cadence or content actually moved the score and the downstream conversions, closing the gap between a tweak and its proof.

1. 1

   Weighting opens like clicks

   An open is the weakest and most polluted signal, yet teams often weight it as heavily as a click. Keep open weight low so the score reflects intent rather than image pre-fetching.

2. 2

   Forgetting to decay the score

   Without decay, a recipient who was active a year ago keeps a high score forever. The score must age so it reflects current intent, not lifetime history.

3. 3

   Never validating against outcomes

   A score that does not correlate with conversion is decoration. Check that high scorers actually convert better than low scorers, and recalibrate the weights when they do not.

4. 4

   Building a black box

   An opaque score nobody can explain is one nobody trusts or acts on. Favour a simple, legible model so the team understands why a recipient sits where they do.

5. 5

   Scoring but not routing

   Calculating a score and then mailing everyone the same thing wastes it. The score earns its value only when it changes what each recipient receives next.