How to find leads on Discord without getting banned

You joined eleven Discord servers in one evening. You set your notification keywords, muted everything else, and waited. Three days later someone in a founder server posts "anyone know a good tool for X?" and you are there, ninety seconds a

You joined eleven Discord servers in one evening. You set your notification keywords, muted everything else, and waited. Three days later someone in a founder server posts "anyone know a good tool for X?" and you are there, ninety seconds after they hit send. You DM them. It goes well. So you do it again the next day, and the day after, and on the fourth morning your account is gone, along with the eleven servers, the notification setup, and the two conversations that were actually going somewhere.

Discord lead generation has a research problem. Almost every guide written about it recommends at least one thing that Discord prohibits in writing. Not frowned upon by the community, not risky in a vague reputational sense: written down, numbered, and attached to an enforcement process. Before you build a motion on this channel, it is worth reading the four documents that govern it, because they rule out most of the tooling category that markets itself at you.

Enforcement being patchy is not the same as the activity being allowed. The account that gets banned is yours, not the vendor's.

Three Community Guidelines rules cover most of the standard playbook

Discord's Community Guidelines are short and numbered, which makes them unusually easy to check your plan against. Three rules matter here.

Rule 13: "Do not send unsolicited bulk messages (or spam) to others." The rule extends past sending, to facilitating, and names the tooling explicitly: spambots, raid tools, account-creation tools, token generators, CAPTCHA-solving services.

Rule 14: "Do not use self-bots or user-bots. Each account must be associated with a human, not a bot." This is the one that quietly kills the most popular DIY approach. A script running on your own user token, watching channels and pinging you, is a self-bot. It does not matter that it only reads and never posts.

Rule 15: "Do not engage with our service in an inauthentic way." Artificially inflating server membership and manipulating engagement metrics both land here.

Discord's Terms of Service sit underneath those and prohibit "scraping our services without our written consent, including by using any robot, spider, crawler, scraper, or other automatic device, process, or software," alongside "auto-messaging" people through the service.

The Developer Policy closes the door the API appears to open

The obvious response to rule 14 is to do it properly: register an app, get a bot token, stop using your user account. That is the right instinct, and it moves you into a different rulebook rather than out of one. The Discord Developer Policy, effective July 8, 2024, is where a lead-generation use case runs out of road.

Rule 5: "Do not contact users on Discord without their explicit permission." The rule names "frequently sending unsolicited direct messages" and messages "not directly related to maintaining or improving an Application's functionality."

Rule 6: "Do not target users with advertisements or marketing."

Rule 7: "Do not contact users outside of Discord without their explicit permission." The body is the part people miss: "You should not contact Discord users outside of the Discord platform using API Data (including any data obtained, disclosed, or inferred through the use of your Application)." That sentence forecloses the enrich-and-email workflow. Spot the intent in Discord, find the person's work email, send a cold email: prohibited, because the trigger was API Data.

Rule 20: "Do not mine or scrape any data, content, or information available on or through Discord services."

Rules 15, 16 and 17 finish the job. API Data may only be used for your app's stated functionality, may not be used to profile users, and may not be disclosed to data brokers or advertising services. A cross-server keyword monitor that builds you a list of people who mentioned your category is, under this policy, profiling users with scraped data for marketing. Every clause is against it.

Server admins are the real gatekeepers, and Discord keeps tightening the gate

A bot only sees a server because an admin installed it. That single fact is what makes Discord structurally different from Reddit or Hacker News, where public content is readable by default. On Discord, monitoring requires an invitation, per server, from a person whose job is keeping vendors out of their community.

Discord has been narrowing the path further. In a change to Privileged Intent access announced in 2026, the review threshold moved from a 100-server count to a 10,000-user count. Apps under 10,000 users can still toggle Message Content, Guild Members and Presence on in the Developer Portal. Past that, you apply, you have 90 days to file, and approved apps now reapply once a year. The application asks what your app does, why it needs each intent, and what your data practices are.

Read that as a product decision rather than a policy detail. Discord is asking every app that reads message content at scale to justify itself annually to a human reviewer. "We surface buying signals for sales teams" is not a use case that survives that conversation, because rules 6 and 16 already say no.

The motion that survives all four documents is unglamorous participation

Here is what is left, and it does work. It just does not automate.

Join servers you have an actual reason to be in and participate for weeks before you need anything. Standing is the whole asset on Discord, and unlike Reddit there is no karma number to proxy it. People remember whether you were useful.

Use Discord's own notification features rather than a script. Keyword notifications and per-channel alerts are built into the client, they run on your account legitimately because a human is driving them, and they cover the "someone mentioned my category" case without a token in sight. This is the boring answer and it is the compliant one.

Reply in the channel, not in DMs. Rule 5 is about unsolicited direct messages, and the public reply is better anyway: it gets read by everyone else with the same problem, and it puts your answer on the record where the server's culture can judge it. The same logic that governs replying on Reddit without getting banned applies here with less margin for error.

Let the other person open the DM. If your public answer was good, some of them will. That inbound DM is explicit permission in the sense rule 5 means it, and the conversation starts from a position where you have already demonstrated the thing you would otherwise have claimed.

Ask the admin before you bring any tooling in. If your product genuinely helps that community, an admin-installed bot is legitimate. It is also one server at a time, which is the honest scaling story.

Discord is a worse first channel than Reddit or Hacker News for most founders

We build a monitoring tool, so this is the section where the incentive would be to tell you Discord is an untapped goldmine. It is not, and the reason is structural rather than competitive.

Reddit and Hacker News are public by default. Anyone can read them, the content is indexed, and the signal is durable enough to act on hours later. Discord is a set of private rooms with doormen. The messages are not indexed, most servers are invite-shaped, and the archive is not yours to keep. The signal-economy framing still holds, but the collection cost per usable signal is several times higher, and the ban risk sits on your personal account instead of a throwaway.

Shadow Inbox is our product and it does not monitor Discord. That is a decision, not a gap in the roadmap. Doing it properly would mean an admin-installed bot per server, which does not compose into a monitoring product, and doing it improperly would mean rule 20. So we watch Reddit and Hacker News, where reading is allowed and the anatomy of a high-intent post is legible without anyone's permission. If a competitor tells you they cover Discord across servers they were never installed in, ask which rule they think that falls under.

Spend the first sixty days of any community-led motion on the public platforms, where Hacker News comments carry the same buying signals without the account risk. Add Discord later, manually, in the three or four servers where you are actually a member. Treat it as a depth channel rather than a volume one, and the rules stop being a problem, because you were not going to break them anyway.

● FAQ

Is it against Discord's rules to look for leads in Discord servers?

Reading public channels in servers you joined normally is not against any rule. What the rules restrict is the machinery people bolt onto that: automated user accounts, scraped message archives, and unsolicited DMs. Discord's Community Guidelines rule 14 bans self-bots outright, and Developer Policy rule 5 bans contacting users without their explicit permission. Participation is fine. Extraction is not.

Can I build a bot that alerts me when someone mentions my keyword?

Yes, if a server admin installs it and you stay inside the Developer Policy. The catch is that the data is fenced: rule 15 limits API Data to your app's stated functionality, and rule 7 says you may not contact Discord users outside Discord using data your app obtained or inferred. So an alert bot can tell you a conversation is happening in a server that invited it. It cannot become a lead list you email later.

Why do so many Discord lead-generation tools exist if this is prohibited?

Because the rules are enforced unevenly and the tools sell well. Developer Policy rule 20 prohibits mining or scraping any data available on or through Discord's services, which is exactly what a cross-server keyword monitor does when no admin has installed it. Enforcement being patchy is not the same as the activity being allowed, and the account that gets banned is yours, not the vendor's.

Does Shadow Inbox monitor Discord?

No, and that is a deliberate call rather than a roadmap gap. Shadow Inbox is our product and it watches Reddit and Hacker News, both of which have public, documented read access. Building the same thing for Discord would mean either an admin-installed bot per server, which does not scale into a monitoring product, or scraping, which rule 20 prohibits. We would rather tell you the channel is hard than sell you a ban.

What is the honest expected volume from Discord as a channel?

Low, and slow to arrive. You are limited to servers that admitted you, you have to actually participate to keep standing, and the reply has to happen in public rather than in a DM. For most founders that is a handful of real conversations per month, not a pipeline. Treat Discord as a depth channel and put your monitoring effort into Reddit and Hacker News first.