Engagement Scoring
Engagement scoring quantifies how actively a lead or account interacts with your brand, content, and product, turning clicks, opens, visits, and quiz completions into a measurable interest signal.
Key takeaways
- Points decay with age, so a score falls on its own when a contact goes quiet.
- The half-life setting trades responsiveness against stability, with no universal value.
- Thresholds must sit above the level passive email opening alone can reach.
- Engagement measures attention, not fit, authority or available budget.
- Off-site research through peers and analyst reports never appears in the score.
In depth
Engagement scoring turns a stream of timestamped behaviours into one number by assigning points per action and then discounting them by age. The decay is what distinguishes it from a simple activity count: a point earned today is worth its full value, the same point earned two months ago only a fraction. Because old points fade, the score falls on its own when someone goes quiet, and that is what lets it be read as a timing signal rather than a lifetime total.
Two settings dominate the result: the point spread between actions and the half-life of the decay. A narrow spread makes newsletter skimmers score like people reading pricing pages; too wide a spread lets a single action decide everything. A short half-life makes the score responsive but jumpy, so contacts fall out of alert range days after a good session, while a long one smooths the curve and keeps stale interest alive. Both trade sensitivity against stability, and neither has a universally correct value.
Scores are used as thresholds attached to an action. Crossing a line notifies an owner, changes a nurture track, or places the contact in a call queue that morning. What matters is that the threshold sits above the level passive behaviour alone can reach, so opening five emails cannot manufacture a sales alert. A completed Pivix quiz is a natural heavy-weight action because it costs the respondent several minutes and produces answers, unlike a click that may have been accidental.
Engagement measures attention, not fit or authority. A curious student and a buying committee member can produce identical scores, which is why it is normally kept beside a separate fit score rather than merged into one number. Activity off your own properties is invisible, so a buyer who researched thoroughly through peers and analyst reports arrives looking cold. And any score can be inflated by an interested person with no budget, so reading a threshold as purchase intent overstates what the data supports.
Example in practice
How to measure it
The test is whether the score separates outcomes. Split contacts into bands at the moment they were scored, then compare how often each band produced a meeting or an opportunity in the following weeks. A working model shows the top band converting several times better than the middle one. If the bands converge, the point weights are not distinguishing genuine intent from reading habit.
Watch the distribution too, not only the average. A model that pushes most contacts into a narrow range cannot rank anything, and a long tail sitting at zero usually means most actions go untracked. Track how many contacts cross the alert threshold per week: a number growing steadily without more traffic is inflation, meaning the threshold rather than the audience has effectively moved.
Common mistakes
The first is scoring everything that can be tracked. Every opened email, every page and every scroll adds points, and after a few months the top of the list fills with subscribers who read attentively but never buy. Score a short list of actions that cost the person effort or reveal intent, and set the rest to zero. A model with six weighted actions usually outperforms one with forty.
The second is setting thresholds once and never revisiting them. Adding a channel or a longer nurture sequence raises everyone's score, the alert threshold stays where it was, and sales starts receiving contacts that would not have qualified last quarter. Review the distribution of scores each quarter and reset the threshold to the percentile you actually intend to route, rather than to a fixed number chosen at launch.
Frequently asked questions
What is recency decay in engagement scoring?
Recency decay lowers the points from an action as time passes, so recent behavior counts more than old behavior. It prevents stale activity from keeping a score artificially high.
How does engagement scoring differ from lead scoring?
Lead scoring usually blends fit and behavior to judge sales-readiness, while engagement scoring isolates behavioral intensity over time. Engagement is often one input into a broader lead or customer score.
Why weight a quiz completion heavily?
Finishing a multi-step quiz signals real intent and gives you qualifying data at once. That makes it a stronger engagement signal than a passive action like an email open, so it deserves more points.
What decay rate should I use for engagement scores?
Set it against your sales cycle rather than accepting a default. If a typical deal takes a quarter, points that halve every few weeks will drop interested buyers out of range before a rep reaches them. A rough rule is a half-life near a third of the cycle, then check whether contacts alert while the interest is still live.
Should engagement and fit be one score or two?
Two, kept side by side. They answer different questions, and merging them lets a highly engaged poor-fit contact reach the same total as a quiet perfect-fit one. A two-axis grid, fit on one side and engagement on the other, produces four cells with genuinely different plays, which a single number cannot represent.
Which actions deserve the most points?
Those that cost the person time or reveal a decision in progress: completing a scorecard quiz, requesting pricing, attending a demo, returning repeatedly to the same product page. Passive receipt of a message deserves very little. The useful test is whether someone uninterested would plausibly do it by accident.
How do I stop scores from inflating over time?
Decay handles most of it, provided it is applied to accumulated points and not only to new ones. Beyond that, cap how much any single action type can contribute, so twenty visits to one page cannot outrank a demo request. Then check the distribution quarterly and set thresholds by percentile instead of a fixed number.
Can engagement scoring work without a marketing automation platform?
Yes, for a small number of actions. A spreadsheet or a simple job reading form submissions and page events can produce a usable score, and the modelling matters more than the tooling. What a platform adds is real-time recalculation and the ability to trigger an action the moment a threshold is crossed.
Why do some highly engaged contacts never buy?
Because engagement reflects interest, which is not the same as authority or budget. Consultants, students and competitors all read attentively. This is exactly why engagement sits next to a fit score, and why qualifying questions on a form or quiz still earn their place: they capture what behaviour cannot reveal alone.