Reminder: Sign Up For ‘AI Ready Communities’
You have just under a month to sign up for our AI Ready Communities Program.
This peer-group, cohort-based program helps your organisation maximise the value of AI within and from communities.
We have 6 places remaining. Click the button below to learn more and sign up.
“What is the point of a community now we’ve implemented AI?”
Over the next few weeks, I’ll outline some of the program’s critical principles to give you a sense of what we’ll cover.
(You can also check out my recent webinar: The Evolution of Enterprise Communities)
On a client stakeholder call recently, I was asked what the value of an enterprise community is when the organisation has launched a new customer support portal that uses RAG (retrieval-augmented generation) to pull data from a variety of official sources and generate a near-perfect answer most of the time.
It’s a great question – and one you will probably also get.
So here’s how to think about the answer (I’ll provide the exact answer at the end of the post).
The Unique Value of Enterprise Communities Has Narrowed (but not diminished)
Three years ago, I wrote that the main purpose of many enterprise communities had narrowed to answering the in-between questions.
These were questions that were too complex to be covered by an FAQ/documentation (an edge case) but didn’t require a member to share private details/personal data (and thus a support rep) to resolve the issue.
This was somewhat due to the introduction of Federated Search.
New tools like Coveo and SearchUnify could retrieve information from a wide variety of sources and display them in a single list.
Community answers weren’t going to rank above official docs (nor should they in many cases).
This didn’t mean communities didn’t still get a lot of questions whose answers were in the docs. They did (and still do – for the time being).
But it does mean it was getting fewer of them and wasn’t providing unique value in the answers.
The introduction of RAG (retrieval augmented generation – essentially creating a complete answer based on the information retrieved) further erodes the need for community to answer these kinds of questions.
Where federated search retrieved a prioritised list of relevant links, RAG creates the complete answer from this information.
A major factor in the decline in engagement in hosted communities is the rapid drop in members asking questions when the answer is already in the organisation’s documentation.
However, it’s worth noting, the community wasn’t providing unique value here anyway – so it’s not a big loss.
But it does change how we think about community value.
Communities Provide Fantastic Value….To Specific Questions!
Here’s the critical question going forward:
→ What is the value organisations can get from community that they can’t get through any other channel?
The answer isn’t generating as many responses as possible.
The answer depends more on the types of knowledge that only the community can answer.
You can see this in the graphic below.
It’s worth going through what each of these things is:
Canonical Knowledge
(what must be true)
This is simply the type of knowledge that must be true. AI must not contradict this kind of knowledge. It’s official information based on facts that shouldn’t vary.
This kind of knowledge should be surfaced in questions such as:
“Is this supported?”
“Does your product allow…?”
“What are the limits for…?”
“Is this compliant with…?”
“Which plan includes…?”
“When does this version go end-of-life?”
“What is your official policy on…?”
AI should rely almost entirely on documentation and policies here. The community should not generally feature in these kinds of answers.
The best use of community here is to flag when this knowledge isn’t accurate or clear, so it can be updated. But that’s usually better enabled by allowing feedback/ratings on the docs themselves.
Procedural/Operational Knowledge
(how things are done)
Procedural knowledge that explains how to do something correctly. This kind of knowledge shows up in response to questions like:
“How do I set up…?”
“What are the steps to configure…?”
“How do I troubleshoot…?”
“How do I migrate from X to Y?”
“What should I check when this fails?”
“How do I integrate with…?”
These questions require a step-by-step execution plan. AI should rely primarily on knowledge bases, runbooks, and support articles here.
Community can play a supporting role when official steps are incomplete, unclear, or lack edge-case details. But, again, enabling feedback on the docs themselves is usually a bigger win.
Experiential Knowledge
(what works in reality)
Now we get to the places where community can deliver far more unique value.
Experiential knowledge is knowledge derived from people who have actually tried the task. It reflects the point at which official guidance is insufficient, and more support is needed. Sometimes this might be a gap in documentation, other times it might simply be unique people trying to do unique things.
This kind of knowledge surfaces in questions like:
“This isn’t working even though I followed the docs”
“Has anyone got this working with…?”
“Does this behave differently when…?”
“What worked for you when…?”
“Is there a workaround for…?”
“Why does this fail only in large environments?”
AI should lean heavily on community answers here, because this is where practical workarounds, hidden constraints, and real-world behaviour are captured.
It’s not uncommon for customer support and success professionals to also rely on community solutions for these edge cases.
Contextual Knowledge
(why things fail or succeed)
This kind of knowledge is where a community really shines.
It’s where this isn’t a single correct answer, but people can understand the trade-offs, constraints, and different approaches to an issue. It’s the knowledge that explains the situation around a problem.
This kind of knowledge shows up in response to questions like:
“What’s the best approach for…?”
“Should I use X or Y?”
“What are the trade-offs between…?”
“Why does this design cause problems at scale?”
“How are others solving…?”
“What should I consider before…?”
These questions require judgment rather than instructions. You can’t write them down because it’s so contextual. There isn’t a single definitive answer – and AI can surface the multiple perspectives and ideas here.
Ephemeral Knowledge
(what’s changing right now)
Now we get to the point where community really outshines other channels in by some margin
Ephemeral knowledge is essentially what’s changing right now. It’s an early warning system. It captures what’s happening right now – even before the docs or the knowledge base has been updated.
This kind of knowledge shows up in response to questions like:
“Is anyone else seeing this issue today?”
“This worked last week – what changed?”
“Did the latest release break…?”
“Is there a known issue with…?”
“Why is this suddenly failing?”
These questions are time-sensitive and require a different approach than any other knowledge type.
Often, they can surface workarounds, solutions, or simply confirmation.
Treat Different Knowledge Types Differently
One principle we will cover in the AI Ready Communities Program is to treat different types of knowledge differently.
At the moment, we separate discussions largely by popularity. The most-viewed discussions receive extra care, but they are typically those that can be answered by existing documentation.
Once we view discussions from the perspective of knowledge completeness, we can begin labelling and improving their hygiene from a unique value perspective.
If you want to learn how to do this, join the program.
A Single Sentence To Explain Unique Community Value
Now reflecting on the original question, a good way of explaining the unique value of community is:
→ “Community is the only place where organisations can see how their product truly behaves in the wild. It exposes edge cases, reveals what actually works, and surfaces what’s changing before the rest of the business even knows.”
Adapt and change as you like – just focus on unique value!
