What we measured scraping a 28,734-member Facebook group

Every tool that scrapes Facebook groups quotes a member count. We wanted to know what an ordinary member can actually reach, so we measured it on a real group of 28,734 people. The answer is 4,240 distinct people, or 14.8%, and the shape of what is missing matters more than the number.

The setup

One group, 28,734 members, joined normally. No special access, no admin rights, no API. Facebook has no public group-members endpoint, so the only surface is what the group UI itself will serve to a logged-in member, which is what we collected.

Three lists, and only two of them work

Facebook exposes members through several different views, and they behave very differently.

The interesting number is not any of those totals. It is the overlap between the first two: 6.4%. Things-in-common and new-joiners are almost disjoint sets, which means harvesting only one of them costs you roughly half the reachable people. Most tooling picks one.

Contributors is worth calling out because it looks like the richest surface and is not. Facebook's own UI reports 7,994 contributors and then serves about a dozen. That is not a scraping failure, it is what the endpoint returns.

The finding that changes what you build

Things-in-common exhausts. New-joiners does not, because it is chronological: new people keep arriving at the top of it.

So a one-time dump of a group is a depreciating asset. It gets you 4,240 people once and then it is done. A weekly check of new joiners costs two to five requests and keeps compounding, indefinitely, for as long as the group grows. Over a year the monitor wins by a wide margin and it is gentler on the platform while doing it.

This is why the Facebook group source is built to be run on a schedule rather than fired once. Deduplication is on the numeric profile id, so a re-run adds only genuinely new people.

What you will not get

Across all 4,240 people: zero email addresses. Not a low number. Zero.

Only 1.6% had a domain anywhere in their bio, which is the one thread you can pull on to reach a company. If your plan for a group was to turn it into an email list, the data does not support it, and no tool changes that because the addresses are not exposed to a member in the first place.

What you do get is names, profile links, bios, categories and follower counts. That is enough to qualify people and reach them on the platform, which is a different motion from cold email and should be planned as one.

Why we published the number

Because 14.8% is the sort of thing you find out three hours into a project, and every tool in this category quotes the group size instead. A page that tells you the ceiling before you start is worth more than one that lets you discover it yourself.

The same measurement lives on the template page, under a heading that says "Limits — read this before you run it", because that is where it is actually useful.