← Articles

Crypto Community Metrics: How to Measure Community Health Beyond Follower Counts

Follower counts are the easiest crypto community metrics to fake and the least useful to report. Retention, active contributors, on-chain behaviour and bot share tell you whether a community is real.

Key takeaways

  • Follower and member counts are unreliable in crypto. X's head of product Nikita Bier said in April 2026 that roughly 80% of crypto activity on the platform may be bots, according to reporting by DigitalToday.

  • Bots are not only a vanity problem. A study of 48 tokens found that coins with more bot posts in their first month generally had lower future returns.

  • A useful scorecard tracks five things: returning members, active contributors, the share of community members who also act on-chain, bot share and how members respond to bad news. Discord, Telegram, X and on-chain tools already give you most of the data.

Crypto community metrics are the numbers a project uses to judge whether its community is healthy, and most teams are still reporting the wrong ones. A Telegram group with 80,000 members or an X account with 200,000 followers looks impressive in a deck. Neither tells you whether anyone would notice if the project went quiet for a month.

At AnkhLabs we treat follower counts as the start of a conversation, not the result. This guide sets out the metrics that do tell you something, how to collect them with tools you already have and how to read them honestly.

Why are follower counts misleading in crypto?

Because they are cheap to inflate and hard to audit. In April 2026 X's head of product Nikita Bier argued that spam replies on crypto accounts are hard to fix with technology alone and that roughly 80% of crypto activity may be bots, DigitalToday reported, citing BeInCrypto. The same report says X removed about 1.7 million spam bot accounts over the previous year and, in January 2026, cut off API access for InfoFi apps, which Bier said were spreading AI-generated content and reply spam.

Whatever the exact share, the direction is clear. A large part of what looks like engagement in crypto is automated, and paid campaigns that reward posting make the problem worse. We covered how those reward-for-posting schemes changed in our piece on Kaito and InfoFi.

There is also evidence that bot-driven attention is a warning sign rather than a neutral number. In a study of 48 cryptocurrencies created between 2019 and 2021, the researchers Khizar Qureshi and Tauhid Zaman found that "cryptocurrencies with more bot posts have lower future returns", using Twitter data from each coin's first month (arXiv). Very low and very high engagement were both associated with weaker performance. Low engagement suggests a lack of interest, and unusually high engagement may point to automation.

What are the crypto community metrics that matter?

A healthy community has people who come back, people who contribute, people who use the product and people who stay when things go wrong. Each of those can be measured.

Metric

What it tells you

Where to get it

Returning members (week 4 and week 12)

Whether people come back after joining

Discord Server Insights, Telegram channel statistics, your own cohort tracking

Active contributors

How many people create value, not just read

Message and thread counts by unique member, forum posts, pull requests, governance proposals

Community-to-chain overlap

Whether the community actually uses the product

Wallets linked through quests or sign-ins, matched against on-chain activity on Dune or your own indexer

Bot share

How much of the activity is fake

Sampling of replies and new joins, account age checks, Sybil analysis on wallets

Sentiment under stress

Whether members defend or abandon the project

Reading channels after unlocks, outages or price drops, support ticket volume

None of these needs an expensive platform. Most of the data already sits inside the channels you run.

How do you measure community retention?

Retention asks a simple question. Of the people who joined in a given week, how many are still active four weeks and twelve weeks later? Track it by cohort, meaning a group who joined at the same time, so that a big campaign does not hide the fact that last month's joiners all left.

The platforms give you a starting point. Telegram says advanced statistics are available for channels with more than 500 subscribers, including growth graphs, post performance, data on who joins, leaves or mutes the channel and how people found it. Discord's Server Insights gives the moderators of Community servers with more than 500 members aggregated, anonymised data about activity on their server. The mute and leave data is the most useful part. A channel that grows by 5,000 and loses 4,000 has a funnel problem, however good the headline looks.

Retention in the community usually predicts retention in the product. Our guide to crypto user retention shows how to connect the two.

How do you count active contributors?

Most members read and never post, and that is normal. The number that matters is how many people regularly create something others use. That includes answering questions, writing guides, moderating, testing features, reporting bugs and taking part in governance.

Count unique contributors per month, not messages. One person posting 400 times is not a community. Then look at concentration. If five people produce most of the useful contributions, the community depends on them and you should know their names. A formal ambassador program can grow that group if it rewards quality rather than volume.

How do you link community activity to on-chain activity?

This is where crypto has an advantage over any other industry. If members connect a wallet, through a quest, a role check or a sign-in, you can see whether the people in your channels also use the product. Match those wallets against on-chain activity in a Dune dashboard or your own data and you get a figure few teams report: the share of your community that actually transacts.

Two cautions apply. Linking a wallet should be optional and clearly explained, because privacy matters to crypto users. And wallets can be farmed as easily as accounts, so filter them before you trust the result. Our article on Sybil detection explains how projects spot one person running many wallets.

The reverse view is just as useful. Of your active on-chain users, how many are in any community channel at all? If the answer is very low, your community may be talking to itself while the real users sit elsewhere.

How do you estimate bot share?

You will not get a perfect number, but a rough one is far better than none. Take a random sample of 100 replies to your posts and 100 recent joiners to your Telegram or Discord each month. Check account age, profile history, whether the reply relates to the post and whether the same text appears across accounts. The share that fails is your working estimate.

Report it next to your growth numbers. A campaign that adds 10,000 followers with a 60% bot share added 4,000 people, and saying so builds more trust with investors and partners than the larger figure. It also tells you which channels and which paid partners to stop using.

What does community health look like under stress?

The best test of a community is a bad week. A token unlock, an exploit on a partner protocol or a sharp price drop shows whether members ask questions, defend the project or leave. Read the channels in the 48 hours after such events and note three things: the share of messages that are questions rather than complaints, how quickly other members answer before the team does, and how many people leave.

A community that answers its own questions in a crisis is worth more than one twice its size that waits for the team. Our guide to crypto community management covers how to build that habit before you need it.

Which numbers should you stop reporting on their own?

Some figures are not useless, but they mislead when they appear without context. Total members and followers are the obvious ones, because they only go up and say nothing about who left in spirit. Impressions are similar. A campaign can produce millions of views from accounts that never visit the product.

Quest completions and message counts are the subtler traps. A completed quest measures a task, not a person, and one farmer can complete hundreds. A busy chat can be a handful of people talking to each other or a flood of generic replies. If you report any of these, put the unique-person figure beside it, along with the share of those people who came back the following month.

The same rule applies to the reports you receive from agencies and partners. Ask what share of the reach came from accounts with real history, and how many of the new members were still active a month later.

What we'd tell a founder

Pick five numbers, measure them the same way every month and show them to your team and investors next to the follower count. Returning members at week four, active contributors, community-to-chain overlap, estimated bot share and leave rate after bad news are a good default set.

Reward behaviour that shows up in those numbers. For Mythical Forest we ran a social farming system with referrals and rewards tied to sustained activity, which reached 1M+ users onboarded. The point of tying rewards to sustained activity is that one-off actions are easy to farm, while coming back every week is much harder to fake. For Degen, the Farcaster-native token and Degen Chain ecosystem, our ongoing campaign has generated 2.5m+ campaign impressions by working inside a community with its own strong culture rather than importing outside tactics.

Be willing to report a smaller, truer number. Investors and exchanges can check for bots themselves, and a project that already shows its bot share looks more credible than one that does not. Our community building and management team sets up this kind of scorecard, and our case studies show the programmes behind it.

If you want a second opinion on how healthy your community really is, talk to AnkhLabs.

The bottom line

Follower counts measure how loud a project has been. The metrics above measure whether anyone is listening and whether they would stay. In a market where a large share of social activity may be automated, the teams that can show real retention, real contributors and real on-chain use will be the ones partners take seriously.

This article is for information only and is not investment advice.



← Articles

Crypto Community Metrics: How to Measure Community Health Beyond Follower Counts

Follower counts are the easiest crypto community metrics to fake and the least useful to report. Retention, active contributors, on-chain behaviour and bot share tell you whether a community is real.

Key takeaways

  • Follower and member counts are unreliable in crypto. X's head of product Nikita Bier said in April 2026 that roughly 80% of crypto activity on the platform may be bots, according to reporting by DigitalToday.

  • Bots are not only a vanity problem. A study of 48 tokens found that coins with more bot posts in their first month generally had lower future returns.

  • A useful scorecard tracks five things: returning members, active contributors, the share of community members who also act on-chain, bot share and how members respond to bad news. Discord, Telegram, X and on-chain tools already give you most of the data.

Crypto community metrics are the numbers a project uses to judge whether its community is healthy, and most teams are still reporting the wrong ones. A Telegram group with 80,000 members or an X account with 200,000 followers looks impressive in a deck. Neither tells you whether anyone would notice if the project went quiet for a month.

At AnkhLabs we treat follower counts as the start of a conversation, not the result. This guide sets out the metrics that do tell you something, how to collect them with tools you already have and how to read them honestly.

Why are follower counts misleading in crypto?

Because they are cheap to inflate and hard to audit. In April 2026 X's head of product Nikita Bier argued that spam replies on crypto accounts are hard to fix with technology alone and that roughly 80% of crypto activity may be bots, DigitalToday reported, citing BeInCrypto. The same report says X removed about 1.7 million spam bot accounts over the previous year and, in January 2026, cut off API access for InfoFi apps, which Bier said were spreading AI-generated content and reply spam.

Whatever the exact share, the direction is clear. A large part of what looks like engagement in crypto is automated, and paid campaigns that reward posting make the problem worse. We covered how those reward-for-posting schemes changed in our piece on Kaito and InfoFi.

There is also evidence that bot-driven attention is a warning sign rather than a neutral number. In a study of 48 cryptocurrencies created between 2019 and 2021, the researchers Khizar Qureshi and Tauhid Zaman found that "cryptocurrencies with more bot posts have lower future returns", using Twitter data from each coin's first month (arXiv). Very low and very high engagement were both associated with weaker performance. Low engagement suggests a lack of interest, and unusually high engagement may point to automation.

What are the crypto community metrics that matter?

A healthy community has people who come back, people who contribute, people who use the product and people who stay when things go wrong. Each of those can be measured.

Metric

What it tells you

Where to get it

Returning members (week 4 and week 12)

Whether people come back after joining

Discord Server Insights, Telegram channel statistics, your own cohort tracking

Active contributors

How many people create value, not just read

Message and thread counts by unique member, forum posts, pull requests, governance proposals

Community-to-chain overlap

Whether the community actually uses the product

Wallets linked through quests or sign-ins, matched against on-chain activity on Dune or your own indexer

Bot share

How much of the activity is fake

Sampling of replies and new joins, account age checks, Sybil analysis on wallets

Sentiment under stress

Whether members defend or abandon the project

Reading channels after unlocks, outages or price drops, support ticket volume

None of these needs an expensive platform. Most of the data already sits inside the channels you run.

How do you measure community retention?

Retention asks a simple question. Of the people who joined in a given week, how many are still active four weeks and twelve weeks later? Track it by cohort, meaning a group who joined at the same time, so that a big campaign does not hide the fact that last month's joiners all left.

The platforms give you a starting point. Telegram says advanced statistics are available for channels with more than 500 subscribers, including growth graphs, post performance, data on who joins, leaves or mutes the channel and how people found it. Discord's Server Insights gives the moderators of Community servers with more than 500 members aggregated, anonymised data about activity on their server. The mute and leave data is the most useful part. A channel that grows by 5,000 and loses 4,000 has a funnel problem, however good the headline looks.

Retention in the community usually predicts retention in the product. Our guide to crypto user retention shows how to connect the two.

How do you count active contributors?

Most members read and never post, and that is normal. The number that matters is how many people regularly create something others use. That includes answering questions, writing guides, moderating, testing features, reporting bugs and taking part in governance.

Count unique contributors per month, not messages. One person posting 400 times is not a community. Then look at concentration. If five people produce most of the useful contributions, the community depends on them and you should know their names. A formal ambassador program can grow that group if it rewards quality rather than volume.

How do you link community activity to on-chain activity?

This is where crypto has an advantage over any other industry. If members connect a wallet, through a quest, a role check or a sign-in, you can see whether the people in your channels also use the product. Match those wallets against on-chain activity in a Dune dashboard or your own data and you get a figure few teams report: the share of your community that actually transacts.

Two cautions apply. Linking a wallet should be optional and clearly explained, because privacy matters to crypto users. And wallets can be farmed as easily as accounts, so filter them before you trust the result. Our article on Sybil detection explains how projects spot one person running many wallets.

The reverse view is just as useful. Of your active on-chain users, how many are in any community channel at all? If the answer is very low, your community may be talking to itself while the real users sit elsewhere.

How do you estimate bot share?

You will not get a perfect number, but a rough one is far better than none. Take a random sample of 100 replies to your posts and 100 recent joiners to your Telegram or Discord each month. Check account age, profile history, whether the reply relates to the post and whether the same text appears across accounts. The share that fails is your working estimate.

Report it next to your growth numbers. A campaign that adds 10,000 followers with a 60% bot share added 4,000 people, and saying so builds more trust with investors and partners than the larger figure. It also tells you which channels and which paid partners to stop using.

What does community health look like under stress?

The best test of a community is a bad week. A token unlock, an exploit on a partner protocol or a sharp price drop shows whether members ask questions, defend the project or leave. Read the channels in the 48 hours after such events and note three things: the share of messages that are questions rather than complaints, how quickly other members answer before the team does, and how many people leave.

A community that answers its own questions in a crisis is worth more than one twice its size that waits for the team. Our guide to crypto community management covers how to build that habit before you need it.

Which numbers should you stop reporting on their own?

Some figures are not useless, but they mislead when they appear without context. Total members and followers are the obvious ones, because they only go up and say nothing about who left in spirit. Impressions are similar. A campaign can produce millions of views from accounts that never visit the product.

Quest completions and message counts are the subtler traps. A completed quest measures a task, not a person, and one farmer can complete hundreds. A busy chat can be a handful of people talking to each other or a flood of generic replies. If you report any of these, put the unique-person figure beside it, along with the share of those people who came back the following month.

The same rule applies to the reports you receive from agencies and partners. Ask what share of the reach came from accounts with real history, and how many of the new members were still active a month later.

What we'd tell a founder

Pick five numbers, measure them the same way every month and show them to your team and investors next to the follower count. Returning members at week four, active contributors, community-to-chain overlap, estimated bot share and leave rate after bad news are a good default set.

Reward behaviour that shows up in those numbers. For Mythical Forest we ran a social farming system with referrals and rewards tied to sustained activity, which reached 1M+ users onboarded. The point of tying rewards to sustained activity is that one-off actions are easy to farm, while coming back every week is much harder to fake. For Degen, the Farcaster-native token and Degen Chain ecosystem, our ongoing campaign has generated 2.5m+ campaign impressions by working inside a community with its own strong culture rather than importing outside tactics.

Be willing to report a smaller, truer number. Investors and exchanges can check for bots themselves, and a project that already shows its bot share looks more credible than one that does not. Our community building and management team sets up this kind of scorecard, and our case studies show the programmes behind it.

If you want a second opinion on how healthy your community really is, talk to AnkhLabs.

The bottom line

Follower counts measure how loud a project has been. The metrics above measure whether anyone is listening and whether they would stay. In a market where a large share of social activity may be automated, the teams that can show real retention, real contributors and real on-chain use will be the ones partners take seriously.

This article is for information only and is not investment advice.