Member engagement: how to actually measure it (not just track logins)
A member who logs in once a year to pay an invoice and a member who logs in weekly score the same on most dashboards: engaged.
That is the problem with login counts. They measure whether someone had a reason to visit, not whether the membership is worth anything to them. And because they are the easiest number to produce, they are the number most organizations end up reporting.
Here is what actually predicts renewal, and how to build something useful from records you already keep.
Start with the question the metric has to answer
The purpose of an engagement measure is not to describe the past. It is to hand you a list of names to act on, early enough to act.
That single requirement rules out most of what gets reported. Renewal rate is a fact about something that already happened. Total event attendance is a fact about the organization, not about a member. Neither produces a name and neither produces it in time.
So the test for any signal below: does it tell you which member, and does it tell you before the renewal notice goes out?
The signals worth having
Event attendance, weighted by recency. The strongest single predictor most organizations have. A member who came to four things last year and nothing since March has already left; the renewal invoice is a formality. Weight recent attendance more heavily than historical, because a member's last six months predict better than their first three years.
Email clicks, not opens. Open rates stopped being trustworthy when Apple's Mail Privacy Protection began pre-fetching images in 2021, which registers an open whether or not anyone looked. Clicks still mean a person did something. Track them, and treat a member who has clicked nothing in ninety days as a signal rather than a statistic.
Whether anyone at the organization has spoken to them. Not a system metric, but the one staff can answer and the one that matters. If nobody on staff or the board can name a person at that member beyond the invoice contact, the membership is institutional, and institutional memberships are the first cut when a budget tightens.
Whether the contact is still current. A membership billed to someone who left in March is a non-renewal that looks like a decision and is actually an undelivered email. This is the most recoverable signal on the list and the most commonly missed.
Dues timeliness, as a trend. Not whether they pay, but whether they pay later each year. A member who used to pay on the day and now pays at the second reminder is telling you something before they tell you.
What to leave out
Logins, for the reason above.
Opens, for the reason above.
Directory profile completeness, unless you know your members actually use the directory. It is easy to measure and easy to mistake for interest, and a member who filled in their profile in 2021 and has done nothing since scores well on it forever.
Anything you cannot pull without an afternoon of work. A signal that requires an export and a spreadsheet is a signal you will use twice and then stop.
Build the crude version first
You do not need a scoring model. You need a list.
Take three questions, answer them per member, and count how many are yes:
- Have they attended anything in the last six months?
- Have they clicked anything in the last ninety days?
- Can someone here name a person there?
Zero out of three, with a renewal inside six months, is your at-risk list. In most chambers it is longer than staff expect and shorter than a scoring model would flag, which is what makes it actionable: it is a list you can work through by phone in a week.
Run it quarterly. The value is not precision. It is that the conversation happens in September rather than at the invoice, which is the difference between a retention call and a save attempt.
When a score is worth having
A weighted score earns its place once the list gets too long to work through by hand, or once you want to see direction rather than state. A member whose score is falling is more interesting than one whose score is low, because low might just be how they have always used the membership.
Two cautions if you build one. Do not let it become a number reported to the board without the names attached, because then it is a metric rather than a tool. And check what it is actually weighting, since a score dominated by whichever signal was easiest to collect will quietly rank your members by how much data you happen to have about them.
How we handle it
Sembr computes an engagement score per member from recorded interaction events, and event attendance and RSVPs are the signals carrying the most weight today.
Two honest limits. The scoring is only as good as what is being recorded, so a chamber running events outside the platform will see a thinner picture. And the at-risk view is deliberately a list of names rather than a chart, because the number was never the point.
That view is on the member portal page, and the September timing argument is in how chambers can use Labour Day to reset and re-engage.