Latest / Community Marketing with Fexingo: Discord, Slack Groups, and Brand Communities / How a 300-Person Discord Replaced a Full Product Research Team
Transcript
- Lucas: So there's this B2B SaaS company — I won't name them because they're still relatively private — but they had been paying a product research agency about a hundred and twenty thousand dollars a year to run user interviews, compile insights, and validate feature ideas. Luna: That's a decent chunk of change for a startup. Lucas: It is. And last year they decided to try something different. They already had a Discord server with about three hundred active members — mostly power users and early adopters. So they turned that server into their primary research engine. And within six months, they dropped the agency contract entirely. Luna: Wait — three hundred people replaced a full research agency? That seems like a bold claim. Lucas: It does. But when you look at how they structured it, it starts to make sense. They didn't just open a #feedback channel and hope for the best. They essentially rebuilt the agency's workflow inside Discord. Luna: Okay, I'm curious. How did they do that? Lucas: If these marketing conversations have sparked something you've actually used, the way these stay ad-free is listener support — buy me a coffee dot com slash fexingo. It's a small way to keep the episodes coming without any sponsor breaks. Luna: Yeah, and it genuinely helps. We don't run ads, so that link is the only thing that keeps the lights on. Back to the Discord — how did they structure it? Lucas: So first, they created dedicated channels for each upcoming feature or experiment. Not a general feedback dump — specific threads for specific questions. For example, they'd post a mockup of a new onboarding flow and say, 'We're testing this with five people tomorrow. Anyone want to hop on a fifteen-minute call?' Luna: And people actually signed up? Lucas: Consistently. They had about a thirty percent opt-in rate from the community. That's roughly ninety members willing to jump on a call whenever needed. Compare that to the agency, which was sourcing maybe eight to ten interviewees per month from a panel. Luna: So speed was a big advantage. Lucas: Massive. They could run a study on Monday and have synthesized insights by Wednesday. The agency took two to three weeks for the same turnaround. But it wasn't just speed — the quality of feedback was different too. Because these were actual users who had context, not paid panelists. Luna: Did they worry about bias though? Power users might not represent the broader user base. Lucas: That's a real risk, and they addressed it by segmenting their community. They had tags for company size, role, and usage frequency. So when they needed feedback from, say, a marketing manager at a fifty-person company, they could ping that specific group. Luna: So they basically built a living research panel. Lucas: Exactly. And the numbers back it up. Over the first six months, the community generated the equivalent of about two full-time researchers' worth of output — based on the number of interviews conducted, surveys completed, and feature validations run. The agency was costing them a hundred and twenty K. The community cost them essentially zero in direct spend, plus maybe ten hours a week from a community manager. Luna: Ten hours a week is not nothing. Lucas: No, but it's a fraction of what they were paying. And the community manager wasn't new — they already had that role. So the incremental cost was minimal. The bigger question is whether this scales. Luna: That's what I was thinking. Three hundred members might work for a company with a few thousand users, but what happens when you have fifty thousand? Does the model break? Lucas: The company we're talking about actually grew to about eight hundred members in the Discord, and they've kept the same model. But they had to add more structure — they created a private beta tester role with a separate channel, and they started using Discord's built-in polls and forms more heavily. Luna: So they still rely on the community, but they're managing the scale with tools. Lucas: Right. And they also started rotating which members get early access to features, so the same people aren't always the ones giving feedback. That reduces fatigue and keeps the sample fresh. Luna: I think the key insight here is that the community replaced the agency, but it didn't replace the research function — it just moved it in-house and made it faster. Lucas: Exactly. The product team still has to analyze the feedback and make decisions. The community is the source, not the analyst. And that's important because some companies hear stories like this and think, 'Great, we'll just open a Discord and get free research.' It doesn't work that way. Luna: What's the biggest mistake companies make when they try this? Lucas: They treat the community as a monolith. They ask vague questions like 'What do you think of our product?' and get a hundred different answers that are impossible to synthesize. The company we studied was very deliberate — every request was specific, time-boxed, and tied to a decision they needed to make. Luna: So structure is everything. Lucas: Structure and respect. If you spam your community with research requests every week, they'll disengage. This company capped requests to two per month per member, and they always closed the loop — they shared what they learned and what changed because of the feedback. Luna: That's smart. People are more willing to participate if they see their input actually matters. Lucas: It's the reciprocity principle. And the result was that their trial to paid conversion rate improved by thirty percent over the year they started doing this. They attributed that directly to building features that better matched what new users actually needed, which they learned from the community. Luna: Thirty percent is huge. Do you think this model could work for a non-SaaS company? Say, a consumer brand? Lucas: I think it can, but the stakes are different. For a consumer brand, community feedback might inform packaging or a new flavor, not a software feature. The turnaround is slower. But the principle — using your most engaged customers as a rapid feedback loop — applies anywhere. Luna: Give me an example. Lucas: There's a direct to consumer pet food company that uses a private Facebook group to test new recipes. They post a concept, get reactions, and only produce the ones that get strong positive signals. They've told me it's cut their product development cycle by about forty percent. Luna: So the same model, different industry. Lucas: Exactly. The key is that the community has to be engaged enough to give you high-quality feedback, and you have to have a system to act on it quickly. If you can't turn feedback into a decision within a week, the community loses momentum. Luna: I think the biggest takeaway for me is that this isn't just about saving money. It's about building a closer relationship between the people who make the product and the people who use it. Lucas: That's the real advantage. The agency was a black box — they'd deliver a report, but the product team never talked to users directly. Now, product managers sit in on community calls every week. They hear the frustration and excitement firsthand. That changes how they prioritize. Luna: And that's something you can't really put a price on. Lucas: No, you can't. But the savings are nice too. A hundred and twenty grand a year for a startup — that could be another engineer or a marketing hire. Luna: So if someone listening wants to try this, what's the first step? Lucas: Don't start by trying to replace an agency. Start by identifying one specific question you need answered — like 'Why are trial users dropping off at step three?' — and see if you can get ten users to talk to you about it. If that works, then build the structure. Start small and prove the model. Luna: That's solid advice. And if it works, you might end up with a community that's not just a marketing channel but your best research lab. Lucas: Exactly. And that's a pretty powerful asset to have.