Here’s a common conundrum:
If you (the community manager) respond to a question in a community, other members are less likely to respond. This makes it harder for top members to earn points and feel a sense of influence.
But if you don’t respond to a question in a community, it can linger and look bad. It also means the person asking a question is waiting for a response and becoming increasingly frustrated.
Most community folks treat this as a binary problem and resort to either answering every question they can or only answering questions after 3 to 4 days have passed. Neither is ideal and reflects a lack of thinking about which questions you should or shouldn’t be answering.
There are two levels to this depending upon what data you have access to.
If you have limited data, you immediately respond to questions which:
- You know it will be hard for most members to answer.
- Don’t have a solution.
- Stress a high level of urgency/frustration.
- Are from first-time participants.
- Are from high-value members/customers.
If you can scrape or analyse the data, you can If you have limited data to see what kind of questions will or won’t be responded to.
You can see an example below:
Product categories 3 and 4 – leave for members
In the above example, you probably don’t need to jump in for questions in several categories (especially product 4 product 3).
Developers and partners – check responses and add value
It’s also clear that developers and partners aren’t getting good responses. So you may want to check the responses you do get and add value where you can.
Product 1 – Escalate after 300 minutes
It’s also clear that the ‘product 1’ category has a high time to first response, but a low response rate. This suggests members are answering the easy questions, but the rest linger. You might set a rule that if a question lingers here for more than 300 minutes you jump in.
Product 2 – Jump in immediately
Finally, product 2 has poor responses all round. You should immediately jump in and answer these questions because the community doesn’t seem to have the expertise to do it.
Like most things, this isn’t a binary problem. You can dive deeper and develop a much better solution.
p.s. If you want to be really fancy, you can build a model using category, subject title, post length, and sender information to predict which questions will receive a response and jump in those that are unlikely to get a reply.
