You have a relevancy problem.
It’s easy to understand, harder to solve.
You launch a community focused around a tight concept (e.g. beach metal detectors instead of metal detecting).
Your first members join and participate because they love beach metal detectors. To them the gap between beach metal detecting and any other kind is as wide as a canyon. Right now, your community is 100% relevant to their unique interests.
As the community grows a more diverse group joins with increasingly diverse questions. This means the overall relevancy of questions begins to decline and the percentage of members who participate declines with it.
You can tackle this if you understand relevancy and the methods to sustain it.
What Is Relevancy To Communities??
Relevancy isn’t binary. An issue (discussion/activity etc..) isn’t either relevant or irrelevant. There are plenty of degrees between the two. Where an issue falls on your relevancy continuum depends upon the impact the issue has and how immediate it feels.
Impact is the perceived weight (or value) of the problem/opportunity/social connection being discussed. Immediacy is how quickly we need to solve it or can take advantage of it etc..
Let’s imagine an update to Discourse terrifyingly breaks our community. We urgently need to fix this problem or risk losing members (and prospective clients!). This makes the problem high impact and high immediacy (HIHI). HIHI issues are those most relevant issue to us.
When members visit your community, you want the majority of your menu of discussions to be HIHI discussions.
But there’s a problem here. A HIHI issue to one member will be very different to another.
What’s most relevant to me isn’t what’s most relevant to you. On a mass community scale, this soon becomes a big problem.
For example, we might ask in our community if anyone else using Discourse knows how to fix the problems caused by the update. While this is the most relevant issue to us, it’s not relevant to the 99% of members using other platforms.
Previously 10 of the previous 10 were relevant. Now you have 9 relevant discussions and 1 irrelevant discussion. Of course 1 irrelevant question doesn’t matter much in the great scheme of things. But when that 1 creeps up to 3, 5, and then 9 (as it inevitably does), you can spot the problem.
If the number of responses to discussions, the level of activity per active members, and the amount of time spent on the site is in decline, you have a big relevancy problem.
The challenge is to keep the experience highly relevant while allowing for growth.
3 Methods To Tackle The Relevancy Problem
Even if you do spot the relevancy problem, you need a method to tackle it.
There are three common methods.
1) Curated Relevancy. This is where you curate a list of the most popular discussions. The assumption is what’s most popular is also most relevant to most people. These tend to be either high impact to a large’ish group of people.
You limit other notifications and send these out as email digests or newsletters. If you can’t do anything else, do a ‘best of’ list and try to find the best content that will be most relevant to most of your members (try not to automate this, that’s lazy).
2) Self-Select Relevancy. This is when you create a list of possible preferences (topics, groups etc…) that members can select from and join. Members then only receive notifications or messages about this content.
The problem is members usually don’t choose at all, they stick with the defaults you give them. To make this work you need to be firm in encouraging members to decide what content they want or which groups they wish to join. They then only receive content about these preferences (an updated level is to let members set up preferences/groups themselves that others can choose from).
3) Automated Relevancy. At the highest level is where you use an algorithm to send people content you suspect will be most relevant to them. This combines what people have viewed so far, what discussions they have participated in and only showing/notifying them of related discussions.
Amazon, Quora, and Facebook use machine learning algorithms to do this well. Each member receives a highly personalised experience.
None of these options are ideal. The first has only a small impact the second is tough to enforce, and the third requires a lot of technical expertise and a customisable platform.
Your solution will probably lie at the outer limit of what your technology (and your boss) lets you do. You might have automated autoresponders based upon member actions which drops them into segmented groups. They then receive unique ‘best of’ newsletters based upon what’s best within that particular topic.
You might run a semantic analysis to identify word trends and set up groups based upon those trends and encourage members to join those groups.
You won’t find a single answer to solve the problem. It’s more important that you’re thinking about how to solve the problem. You need to explore what your technology can and can’t do here and develop the best solution you can with what you have.
