Contact engagement score

A contact engagement score is a single weighted number that summarises how actively one contact is interacting across the channels you can measure. Instead of reading email opens, login frequency, meeting attendance, and feature usage as separate streams, the score combines them into one figure you can rank and compare. A contact who logs in daily, opens every email, and attended the last two calls scores high. A contact who has gone quiet for a month scores low.

The score works by assigning a weight to each type of signal and a recency decay so that recent activity counts more than old activity. A product login this week matters more than one from two months ago. A demo attendance is worth more than an email open. The weighting is what turns raw activity into a measure of intent and relationship health rather than a noisy count of clicks.

Contact engagement score matters because it makes a large book of contacts triageable. A customer success manager with 200 accounts cannot read every signal on every contact. The score surfaces the handful that are cooling off and the handful that are heating up, so attention goes where it changes an outcome. Treated well, it is an early warning for churn and an early signal of expansion or buying intent.

The score is a weighted sum of engagement signals, each adjusted for how recently it happened. The arithmetic is straightforward. The judgement is in choosing the signals, setting the weights, and tuning the decay so the number reflects real engagement rather than whichever channel is easiest to count.

1. 1

   Choose the signals

   Pick the actions that genuinely indicate engagement for your business. Common choices are product logins, feature usage, email opens and clicks, content downloads, meeting attendance, and replies. Avoid vanity signals that are easy to trigger but say little about real intent.

2. 2

   Assign a weight to each signal

   Weight signals by how strongly they predict the outcome you care about, such as renewal or a closed deal. A reply or a meeting attended should outweigh an email open. Keep weights explainable so the team trusts the score and can challenge it.

3. 3

   Apply a recency decay

   Multiply each signal by a recency factor so activity fades as it ages. A common approach halves the contribution of a signal every set number of days. Without decay, a contact who was active last quarter looks identical to one active this week.

4. 4

   Sum and normalise

   Add the weighted, decayed contributions into one raw score, then normalise to a fixed range such as 0 to 100 so scores are comparable across contacts and over time. Normalising also keeps a single hyperactive channel from dominating the total.

A worked example shows how the pieces combine. Suppose logins carry a weight of 5, email opens a weight of 1, and meetings attended a weight of 10, with a recency decay that halves a signal every 14 days. A contact with four recent logins, six recent opens, and one meeting two weeks ago scores roughly (4 x 5) + (6 x 1) + (1 x 10 x 0.5), which is 31 before normalising. Change the meeting to last week and the same contact jumps to 36. The recency factor is doing real work, which is exactly the point.

A metric tree decomposes the engagement score into the channels and behaviours that feed it, so a falling score points to a specific cause rather than a vague sense that a contact is slipping. This is the difference between a dashboard that shows the score dropped and a decision about which team reaches out and how.

The first level splits the score by engagement type: product usage, email and content, human interactions, and intent signals. Each branch decomposes further. Product usage breaks into login frequency, depth of feature use, and time in app. Human interactions break into meetings, replies, and support contacts. When the score moves, the tree shows which branch moved, and each branch maps to a different play.

This structure makes the score actionable instead of decorative. A score that fell because product logins collapsed is a product-adoption problem for customer success. The same score drop driven by ignored emails is a messaging problem for marketing. Without the decomposition, every declining score looks identical and the response is a generic check-in that often misses the real issue.

Because the score is built from your own weights and scale, there is no universal benchmark number. What benchmarks well is the band a contact sits in and how that band predicts outcomes. The table below uses a normalised 0 to 100 scale and describes how each band typically behaves so you can calibrate your own thresholds.

The most reliable way to set these bands is to back-test against outcomes. Look at contacts who renewed or closed and find the score range they sat in beforehand. Look at the ones who churned or went cold. Let the data set the thresholds rather than guessing, then revisit them as the model and the business change.

Improving the score means lifting genuine engagement, not gaming the easiest signal. The risk with any score is that teams optimise the number rather than the relationship. The work is to raise the engagement the score is meant to measure and to keep the score itself honest.

The metric tree approach to engagement starts by identifying which branch is driving a contacts movement, then routes the response to the owner of that branch. KPI Tree lets you connect each engagement driver to its accountable owner, so a collapse in product usage reaches the customer success manager and a fall in email engagement reaches marketing. When a score crosses a threshold, the platform pushes the change to the accountable owner rather than waiting for a weekly review, and the verified impact loop checks whether the outreach actually lifted the score. That closes the gap between seeing a contact cool off and doing something about it in time.

1. 1

   Overweighting easy signals

   Email opens and page views are cheap to trigger and weakly tied to outcomes. Leaning on them inflates scores for contacts who are not really engaged and crowds out the signals that actually predict renewal or purchase.

2. 2

   Forgetting recency decay

   Without a decay, a burst of activity from last quarter keeps a contact scoring high long after they went quiet. The score then describes the past, not the present, which is the opposite of what an early warning should do.

3. 3

   Treating the score as truth

   The score is a proxy. A low number can mean a quiet but loyal buyer, and a high number can be one noisy channel. Use it to prioritise attention, then confirm with a human read before acting on a renewal or a deal.

4. 4

   Tracking the number without the composition

   Two contacts with the same score can be completely different, one built on product use and meetings, the other on opens alone. Reading only the headline number hides which engagement is real and which is fragile.

5. 5

   Never recalibrating the model

   Weights set once and left untouched drift out of line with how the business actually wins and loses. A score that no longer predicts outcomes gets quietly ignored by the people it was built to help.