Predicting Forum Value Is Becoming Harder
In the past few months, several organisations have asked us to appraise the value of their forum.
To do this, we usually need two things.
A measure to estimate community value (support deflection, retention, brand preference, etc).
A means of measuring the direction of travel (i.e., is that value increasing or decreasing over time?).
The second part of this has suddenly become infinitely less predictable.
A forum may deliver a given value in support of deflection or retention, but you can no longer predict future value through sheer linear extrapolation (i.e., multiplying present value by [x] months/years, etc.)
This is primarily due to LLMs (large language models), which can detect and connect patterns across vast quantities of semi-structured information in forums, generating value from forums many multiples higher than previously delivered.
This means on a traditional measure, the value might be declining. But when factoring in AI, the value might rise exponentially.
The future increasingly looks less dependent on humans navigating forums directly, but LLMs serving as the interface between users and the sum of knowledge across internal systems – particularly forum knowledge.
What Do Forums Offer That No Other Channel Provides?
Forums provide value that no other channel can easily match.
Forums contain:
Real language. It shows how members describe issues and connect them to the solutions that work (and those that don’t).
Edge cases, troubleshooting, and workarounds. Forums collect countless edge case solutions, collaborative troubleshooting, and workarounds (which sometimes even staff members won’t propose).
Long-tail questions. It’s simply not possible for organisations to create and document answers to the thousands, if not tens of thousands, of questions many customers will supply. Forums help provide that answer.
Multi-turn reasoning. Forums offer the kind of multi-turn reasoning that other channels don’t provide. I.e., a member presents the problem, poses questions, provides answers, and offers clarification, etc.
It’s precisely this kind of content which can be combined with all the other organisations’ IP (and that on social media to feed and create:
AI support assistants.
Product onboarding bots.
RAG (retrieval-augmented generation) search systems.
“Ask the community” AI widgets.
Now it’s possible that you can get this data from other channels – potentially Reddit, StackOverflow, or social media.
But it’s also clear that these channels are increasingly likely to charge a license fee to access their data.
This means that even if engagement is declining or the forum is inactive, now might be a terrible time to close it. LLMs are just about to turn all the knowledge in that forum into gold.
The Real Risk Isn’t Engagement Decline – It’s Data Loss
The absolute risk of closing a forum today isn’t a decline; it’s the loss of data. The moment an organisation kills its forum, it loses:
SEO value
Internal knowledge for employees.
AI-trainable support corpus.
Ability to build future tools (without licensing other channels).
Imagine deleting 15 years of problem-solving because it looks inactive or engagement is sliding.
That would be a tragedy.
Do You Want To Delete Your Knowledge Goldmine?
If we think about the new value of forums, it’s shifted from the old to the latest.
The previous value of forums was to serve as an engaging channel for building member relationships, facilitating discussions, and, primarily, deflecting support tickets.
The new value is to serve as a corpus for AI support and search. It fills the natural gaps in knowledge that can’t be gathered just by reviewing the
You can see this below:
So, What Does All This Mean?
Forums are oil fields. For years, we focused only on surface oil puddles, engagement, replies, and support deflection. But LLMs are the refinery. Now the crude beneath the ground becomes priceless.
LLMs are only just beginning to use forums to deliver incredible value to our members. It would be a tragedy to close down a mature forum now.
Whilst forums today are more challenging to start (due to changing preferences and declining traffic), they have become infinitely more valuable.
The real value of forums over the long term is less about activity and more about the knowledge they provide, enabling LLMs to build a more complete dataset and deliver better results to members.
