Discord Mattermost migration FAQ

On March 1th 2026 we migrate our KoLab community from Discord to a self-hosted Mattermost instance. This document aims to answer the How and Why.

Why leave Discord?

Discord has served KoLab well for many years. As a large common denominator it has been a familiar platform to a large audience. But some developments have been troubling. In particular its handling of personal data. Especially now that it has committed to rolling out universal ID and age verification, it will be collecting vast amounts of personal data and already there have been data breaches.

Not wanting to get their faces scanned or ID documents uploaded to a foreign company in order to communicate with friends over the internet has been the straw that broke the camel's back for some KoLab members, and accelerated the search for an alternative where we reclaim full ownership of our own data on our own terms.

For several months we ran an evaluation of a number of alternative platforms and ultimately decided on self-hosting a Mattermost server.

Where's your Mattermost server?

Self-hosted on a European VPS and open for all to join at: https://mattermost.ko-lab.space/signup_user_complete/?id=nk7dg5o8tjdqpxdu16i5ic3i5o

Did you kill off your Discord server?

No. It's still there for people to find us, search history and even see a real time feed of all channels mirrored and bridged from Mattermost. However, the Discord server has been set to read-only and pinned announcements are in place with instructions on how to join the Mattermost server.

Aren't you gonna lose a lot of user this way?

We hope not!

Discord having such a massive user base makes it easy for existing Discord users to join another Discord server without having to create another account. This is a great way to grow a community and something that will be harder on a self-hosted service that does require users to create a new account. That said, the trickle of existing Discord users that happened upon our server never had great conversion rates towards long term active users. Whereas users that found KoLab elsewhere and took more effort to reach us, consistently have higher conversion rates. We will likely lose the long tail of inactive users, but hope to retain those that actively shape our community.

Isn't Mattermost actively enshittifying their product?

In a sense, yeah.

Mattermost carved out a place for itself by catering to self-hosting and open source communities, offering an open source alternative to the highly successful commercial platform Slack. On its way towards growth and profitability, it followed the Open Core business model, funded through venture capital. With that comes an obligation to forever increase revenue. Then when growth falters, this often leads companies to cull their open source versions and move features into their commercial, paid tiers in order to coerce more users and organisations to becoming paying customers.

This is pattern is commonly known as enshittification.

In Mattermost's model, a lot of features are deliberately not implemented as part of their open source core, but instead as separate plugins that are not open source. Some of these are preinstalled, others are need to be installed separately. And where some of these can be used without a license, many cannot.

Over the past few years, Mattermost has made several steps to actively remove functionality from their open source version and moving it to their paid offerings. Examples include:

• Removed Single SignOn and external Identity providers
• Removed LDAP / Active Directory support
• Removed Playbooks (workflow automation)
• In late 2025, implemented a 10,000 message limit on the free "Entry Edition" version
• Pulled group audio call support from the free tiers
• Removed groups, guest users and access controls

Charging for some features in order to sustain and grow the business is part of the Open Core model and typically companies target features naturally aimed at businesses and large enterprises. SSO, LDAP, SAML, High-Availability, Auditing, etc, are often features small communities and individuals can do without while still retaining a functional product.

Today though, some would argue that in recent times, Mattermost has been overly aggressive and deliberately crippling their open source core. It also no longer mentions the open source "Team Edition" version on its product page, instead steering new users exclusively towards the commercial versions, while burying "Team Edition" instructions.

So then,.. why?

The market for online communication platforms is very fragmented and there is nothing that pleases everyone. In a diverse group of people, the complicated, multi-dimensional Venn diagram of wants, needs, ideology, preferences, tastes, etc, leaves very little overlap.

After several months of evaluating a number of different platform, Mattermost Team Edition (the actual fully open source version) still came up as the the least bad fit.

As a small community with modest needs we feel the _current_ feature set of Mattermost Team Edition to be sufficient and rely either on the company to limit further enshittification, or for a future fork (is "LibreMost" available still?) to carry the torch. Actually, there's already MatterMostly.

Mattermost is terrible, why not Slack?

See So then, why?

But also, it's another commercial cloud offering with usage caps and limited features on the free plan. It being a cloud offering, you don't really own your data and when Salesforce cuts off access to your own data after 90 days in the free tier, we can assume the company will retain your data for its own current of future use.

Mattermost is terrible, why not Matrix?

See So then, why?

But also, it's arguably the least user/normy friendly of the bunch.

Mattermost is terrible, why not Nextcloud Talk?

It's a fair option. Nextcloud Talk also does audio and video calls without arbitrary limits and it comes with Nextcloud itself, which is also useful.

It is a bit of a resource hog to host though. Is this inherent to the php stack?

Mattermost is terrible, why not IRC?

The OG group chat experience.

You can question IRC’s suitability for KoLab when one of our requirements is that the barrier to entry should be as low as possible. IRC has a lot of concepts and characteristics that you don’t find in more modern chat platforms:

• NickServ
• DCC
• No standardization around:
  • Text formatting / markdown
  • Inline rendering of images, URL cards, videos, etc.
  • Notifications / mentions
  • Fine-grained ACLs / groups
• Less privacy-focused (e.g., IPs visible in /whois)
• No message threads
• No native audio calls
• No native screen sharing
• Awkward channel discovery (unless you’re already familiar with /list and other IRC lingo)
• Inconsistent experiences across different clients because much functionality is left to the client

IRC is very old, and its “web 1.0” roots are clearly visible. Since then, chat platforms have evolved into what they are today. The majority of current members did not grow up using IRC as their daily driver, and a not insignificant portion would probably be seeing it for the first time (ratios that will only increase as the audience gets younger).

Mattermost is terrible, why not Discourse?

Discourse is really a forum, rather than a chat system.

Mattermost is terrible, why not Zulip?

Also a fair contender. We test-drove Zulip for several months and liked its "named topic" concept, but ultimately preferred Mattermost's more conventional approach. Also a bit of a resource hog to host. Is this inherent to php?