Because the membership number and the attendance number are measuring two entirely different things, and only one of them has to be earned every single time. Your roster grows through a distribution list that nobody ever leaves. Your room count depends on a person deciding, on a Tuesday, with three deadlines and a school pickup, that this hour is worth it. A list is inventory. A room is a choice.
So the pattern you are seeing is not usually a decline in interest. It is the slow separation of a static number that only goes up from a dynamic number that has to be re-won each quarter. The membership figure was probably never a measure of engagement, and once you say that out loud, the falling attendance stops being a mystery and starts being a set of specific, fixable habits.
Below are the six that show up most often, and the metrics worth swapping in.
Roster, subscriber, member and participant are four different numbers
Most women's networks report one figure and quietly mean four. Separating them takes an afternoon and changes every conversation afterward.
| Number | What it actually is | Typical direction |
|---|---|---|
| Roster | Everyone ever added to the group record, including leavers not yet removed | Only rises |
| Subscriber | Active employees on the mailing list or Teams channel today | Rises slowly |
| Member | People who took a deliberate join action in the last twelve months | Flat or gently down |
| Participant | Attended at least one event or gave one survey response in a quarter | Moves with programming |
Run the arithmetic once. A group reporting 840 members might have 840 on the roster, 690 still employed and subscribed, 310 who took a join action in the past year, and 96 who attended something last quarter. Every one of those numbers is honest. Only the last one tells you whether the program works.
The leaver problem
Nobody owns roster hygiene. HRIS terminations flow to payroll and to IT, but rarely to a volunteer-maintained ERG list. If your company runs seasonal turnover or completed a reduction, a year of untouched roster can hold a meaningful share of people who no longer badge in. Reconcile against active headcount quarterly, not annually.
Keep reading: Where is US corporate DEI reporting heading, and what should our ERG be documenting now?
Counting people who never opted out as engaged
Auto-enrollment feels generous. Add every woman at the company to the network, no friction, instant scale. It also destroys your ability to measure anything, because you have replaced a signal of interest with a signal of demographics.
The tell is a wide gap between open rate and attendance. If your newsletter opens at a healthy clip but events draw a fraction of that, some of those opens are people scanning to confirm nothing needs their attention. That is not engagement, it is triage.
The fix is not to shrink the list. Keep broad reach for announcements, but define membership as an affirmative act: a join click, an RSVP, a mentorship signup, a chapter form. Then report both. Reach of 900 with 280 affirmed members is a far more credible slide than 900 members with no definition attached.
Scheduling against shift patterns, time zones and caregiving hours
Attendance decline is often a calendar problem wearing a motivation costume. Walk your last eight events and write the start time next to the audience you were trying to reach.
- Noon Eastern is 9am on the West Coast, which lands during morning standups and school drop-off.
- 4pm or 5pm local is the hardest hour of the day for anyone with a daycare pickup window and a late fee.
- Manufacturing, retail, clinical and contact center staff cannot leave a floor or a phone queue for a midday webinar. If your network spans exempt and non-exempt populations, a single time slot excludes one of them by design.
- Non-exempt attendance is compensable time under the Fair Labor Standards Act when it is during work hours or effectively expected. If nobody has told frontline managers that ERG attendance is paid, those managers will decline the release, and you will read it as apathy.
Rotate slots deliberately across a year and log the slot with the attendance. Two cycles of that data will tell you more than any engagement survey.
Keep reading: What actually happens in the two weeks before an ERG annual impact report goes to the board?
Programming for the same fifty people every cycle
Every network develops a core. They show up, they volunteer, they give great feedback, and their feedback quietly steers programming toward what the core already likes. Repeat that for three years and you have a well-run club with a shrinking addressable audience.
A simple diagnostic: pull last year's attendance and count distinct attendees against total attendance. If you ran 12 events with 40 average attendance, that is 480 seats. If those seats were filled by 62 distinct people, your average person attended nearly eight times. Impressive loyalty, narrow reach. If the same 480 seats came from 300 distinct people, you have breadth but low repeat, which is a different problem and needs a different fix.
Two numbers worth tracking forever
- First-time attendee share: of the people in the room, how many had never attended before. Below roughly one in six, the program has stopped recruiting.
- Return rate: of first-timers from two quarters ago, how many came back. This is the honest measure of whether the experience was worth the hour.
Treating the annual survey as the only listening channel
An annual pulse arrives too late to change anything and asks people to remember eleven months of feeling. It also over-samples the core, because the core is who reads the email asking them to take it.
Replace the single instrument with a layered one. A two-question post-event pulse sent within 24 hours, one number and one open text box. A short quarterly read on a rotating theme. The annual instrument kept for trend continuity only. The post-event pulse is where the useful material lives, because it captures the person who came once and did not return, and that person never answers the annual survey.
Watch for the participation bias in your own results. If the annual survey is answered mostly by attendees, you are measuring the satisfaction of people already satisfied, and calling it program health.
See how ERGCircle handles this for employee resource group programs
Losing a chapter when its lead leaves
Site chapters run on one person more often than anyone admits. When she is promoted, relocates or steps back, the chapter usually does not announce a closure. It simply stops appearing on the calendar, and the roster keeps counting its members for another year.
Look for chapters with zero events in two consecutive quarters. That is a dormant chapter, and it belongs in a separate line on your report rather than dissolved into a headline total. Naming it protects your credibility and, more usefully, it creates the case for the succession practice that prevents the next one: named co-leads with staggered terms, a shared calendar and budget owner, and a handover checklist that includes the vendor contacts and the room booking access.
The metrics to swap in and how to restate history honestly
Here is a replacement set that survives scrutiny.
| Retire | Adopt | Why it holds up |
|---|---|---|
| Total members | Active participants, trailing 12 months | Tied to an action, not a list |
| Event count | Distinct attendees and first-time share | Measures reach, not activity |
| Survey score | Post-event pulse with response rate shown | Exposes its own sample bias |
| Chapters | Active chapters, with dormant listed separately | Prevents phantom footprint |
| Budget spent | Cost per distinct participant | Speaks the language of the finance review |
Restating history is the part people dread, and it is easier than expected if you do it in one move rather than by degrees. Present both series side by side for the prior two years, label the old one clearly as roster-based, label the new one as participation-based, and state the definition change in a single sentence at the top of the slide. Nobody has ever been penalized for tightening a definition in public. People do get penalized for a number that quietly changes meaning between two reviews.
Expect the new participation figure to look smaller. That is the point. A smaller number you can defend line by line will get you a budget conversation. A larger number that dissolves under a follow-up question will not.
Where to start this quarter
Pick one event, capture attendance properly with names rather than a headcount, send the two-question pulse the next morning, and reconcile the roster against active headcount. That single cycle gives you distinct attendees, first-time share, a real response rate and a clean denominator. Everything else in this article builds on those four.
Keeping that record consistently across chapters and quarters is the hard part, which is exactly what ERGCircle is built for: attendance captured per event, membership tied to an affirmative join, survey pulses attached to the event that prompted them, and budget divided by the participants you can actually name. When the board asks how the number is defined, the definition is already in the report.
