Community-led growth: the honest ten-year version

Every second pitch deck from 2021 to 2023 contained the word "community" next to the word "moat". A Slack workspace would be spun up, a launch post would get forty members, and by month four the workspace would contain three people: a founder, a community manager writing prompts into the void, and someone who joined to sell crypto.

We have helped design, name, launch and — twice — deliberately close B2B communities. Some became genuine growth engines. One became an expensive group chat. This is the honest version: what community actually costs, when it's worth it, and how to measure a channel that refuses to show up in your attribution dashboard.

First, a definition that excludes most things

A community is not an audience. An audience listens; a community talks to itself. The test we use: if your team stopped posting tomorrow, would anything happen next week? If yes — members answering each other's questions, organising their own events, arguing about best practice — you have a community. If the room goes silent, you have a newsletter with a chat room attached.

This distinction matters because the economics are completely different. An audience scales with content output. A community scales with the density and quality of member-to-member relationships, which you can seed and nurture but never manufacture. Confusing the two is how companies end up paying a salary for an empty room.

The bill, itemised

Nobody puts the real cost in the deck. Here it is.

People. A healthy B2B community needs at least one person spending meaningful hours inside it — welcoming, connecting, moderating, programming. Part-time attention produces part-time communities. For anything beyond a few hundred members, budget 0.5–1.0 FTE, and choose someone with genuine standing in the domain, not an intern with a Canva login. Members can smell delegated enthusiasm instantly.

Time. Communities compound on a timescale that makes CFOs uncomfortable. The ones that worked took eighteen to thirty months to reach the point where member-to-member value exceeded staff-to-member effort. If your planning horizon is two quarters, this is not your channel. Buy ads instead; at least their decay is honest.

Consistency. The cadence is the product. Weekly office hours that happen 47 weeks a year build a habit; office hours that happen "when we can" build nothing. When we helped Hearthbrew shape their subscriber community, the single most load-bearing decision was a boring one: the roast-along livestream happens every second Thursday, in perpetuity, hosted by a real roaster, and it has never been cancelled for a marketing opportunity. Rituals beat campaigns.

Taste. Moderation is editorial work. Deciding what behaviour is welcome, which self-promotion is fine and which is spam, when to let an argument run and when to step in — these are judgement calls made daily, and they set the culture more than any welcome doc. Under-moderated spaces die of noise; over-moderated ones die of fear.

When we say yes: the three conditions

We recommend building a community only when all three of these hold.

1. Your customers have a durable, ongoing practice — not a one-off task. Operators, developers, finance teams, growers, clinicians: people who do the thing every week and will still be doing it in five years. A community of people learning to file their first BAS statement has a built-in expiry date. A community of bookkeepers does not — which is why Northwind Ledger invested in practitioner roundtables rather than a public forum.

2. Members get more value from each other than from you. Peer benchmarking ("what are other councils spending per tonne?"), war stories, templates, hiring. If the honest answer is "they mostly want us to answer their questions," build excellent support and a great knowledge base. Those are noble things. They are not a community.

3. Someone senior will show up as themselves. Communities are built by named humans, not brand accounts. If no founder, principal or lead practitioner will be visibly, regularly present, the signal to members is clear: this room is a marketing channel with chairs.

Miss any one of these and we will talk you out of it. We have talked clients out of it more often than into it, and both clients who took the advice are still speaking to us.

Formats that sustain (and one that doesn't)

After a decade, the durable formats are unglamorous:

• Office hours. A recurring video call where a practitioner works through member questions live. Low production, high trust, and every session generates a month of quotable, SEO-able material. The crowd stays small and that's fine — thirty regulars who would miss it beats three hundred lurkers.
• Member showcases. Monthly sessions where a member walks through their actual work — a dashboard, a budget model, a grow log — warts included. This does three jobs at once: recognition for the presenter, concrete learning for everyone else, and proof to prospects that your customers are the kind of people they want to be near.
• A jobs and intros channel. Unglamorous, perpetually active, and the single best retention mechanic we've measured. People who found a hire or a job through a community do not churn from it.
• Small, closed cohorts. Eight to twelve people, one domain, a facilitated call every fortnight. Expensive per member, transformative per relationship. This is where the darkest of dark-funnel value lives.

The format that consistently disappoints: the big public forum. Asynchronous, high-noise, and punished twice — once by spam, once by Google, since thin forum threads can actively drag a domain's perceived quality. If you're not prepared to prune it like our content team prunes a bloated blog (see our content pruning playbook for the mindset), don't plant it.

Measuring the dark funnel without lying to yourself

Community influence happens in places analytics can't see: Slack threads, podcast mentions, "a colleague sent me this". Our piece on attribution covers the general philosophy; here's the community-specific toolkit.

Self-reported attribution, asked properly. Add "How did you hear about us?" to signup and demo forms as a free-text field, not a dropdown. Read the answers monthly. When "someone in the Slack group mentioned you" shows up repeatedly, that's signal. At one client it went from 4% of answers to 31% over two years — invisible in GA4, obvious in a spreadsheet.

Community-cohort tracking. Tag members in your CRM and compare them to matched non-members: close rate, sales-cycle length, expansion revenue, retention. You cannot fully deconfound this — people who join communities are already more engaged — so report it as a range, not a point estimate. "Members retain somewhere between 1.3 and 1.8 times better" is an honest sentence.

Citation counting. Community content gets quoted in places that do show up: decks, RFPs, competitor battle cards. Ask sales to log when a community artefact surfaces in a deal. It turns "vibes" into a countable thing.

What we don't do: last-click attribution with a promo code. Communities are not coupon channels, and measuring them like one is how they get defunded in the quarter they were about to compound.

Communities we would not build

For a pre-product-market-fit startup (the community will be you talking to yourself about a product that keeps changing). For a commodity purchase nobody discusses twice. As a hostage wall in front of support answers. As a "growth hack" bolted onto a churn problem — members leaving your product will not be retained by a Discord. And for any organisation where the honest maintenance budget is zero after launch: an abandoned community is worse than none, because it's a public monument to a promise you stopped keeping.

The pattern underneath all our growth work — from lifecycle email to the way we structure engagements — is the same: compound assets beat rented attention, but compound assets demand rent of their own. Community is the highest-rent, highest-yield property in the portfolio. Move in only if you plan to stay.

Key takeaways

• A community talks to itself when you're not in the room; an audience doesn't. Know which one you're actually building.
• Budget the real cost: 0.5–1.0 FTE, eighteen-plus months of patience, a cadence that never slips, and daily moderation judgement.
• Say yes only when customers share a durable practice, value each other more than they value you, and a senior human will visibly show up.
• Office hours, member showcases, jobs channels and small cohorts sustain; big public forums mostly don't.
• Measure with self-reported attribution, cohort comparisons reported as ranges, and sales-logged citations — never with promo codes.
• An abandoned community is a public monument to a broken promise. Closing one deliberately is a respectable outcome.

FAQ

How big should a B2B community be before it's "working"?

Far smaller than decks suggest. Fifty genuinely active members who answer each other is a working community. We use a health ratio: at least 30% of monthly active participants posting or replying, and at least half of questions answered by non-staff within a day. When those hold, size is vanity.

Slack, Discord, or a proper platform?

The tool matters less than the rituals and the moderation. Choose where your members already live. Slack for operators who live there during work hours; Discord for developer-adjacent and consumer-skewing crowds; a forum platform only if you genuinely need searchable, indexable, long-lived threads and can staff the pruning. Migrating later is painful but survivable; migrating a dead community is pointless.

Should the community live on our domain?

Split it. Conversation lives where conversation is easy (Slack, Discord, in-product). Durable artefacts — showcase recordings, templates, benchmarks — should live on your domain, indexable and owned. That's the part that pays SEO dividends and survives platform decay.

How do we close a community without burning trust?

Announce early, explain plainly, export the artefacts, and redirect the energy somewhere real — a mailing list, quarterly meetups, an open Slack channel run by a member rather than the brand. We've done this twice; both times, members respected the honesty more than they'd enjoyed the declining room.

Can AI moderators run a community?

They can triage spam and surface unanswered questions. They cannot replace presence — the named human who remembers your situation from last month is the entire reason members stay. Use automation for hygiene; never for relationship.