Once, when I was working as a community manager, I got into a very difficult situation in a community chat, after which I did not make any important decisions about the community without looking at the data.
The situation was that I saw one user clearly picking on another user’s words and trying to steer the discussion towards a showdown. As a community manager, I decided to intervene in the dialogue and bring it to something constructive. I was prepared for the “bully” to turn on me. What I was not prepared for was that another community member would support the “bully”. One of them was telling tall tales, and the other “validated” everything the first one said. Soon, more users were involved in the squabble and the dialogue turned into a real drama. The situation was resolved not through arguments and reasoning, but only after we discovered that the “bully” and their “supporters” were one person with several accounts. If I had not had access to the data, the situation would have ended much worse for the community.
This is just one of many stories where trusting my “community management instincts” has gotten me into trouble. Like any other community manager, I’ve thought to myself many times that it would be great to find a troll before they cause harm, or to run a community initiative before users start leaving the platform, etc. Today, I clearly understand that all of this is possible, all we need is to use the data we have correctly.
I think most people reading this post will agree with me that using data to make decisions about communities is very important. At the same time, in reality, few people do it. Why? I think the reason is that working with data takes a lot of time, is difficult, and requires serious technical skills, which not everyone has. This is the essence of the problem that I started working on last fall.
I want to empower community managers to make better, data-driven decisions.
How? In short, by making data analysis accessible and understandable to every community manager.
In case you want to know more detail, the plan consists of three steps:
1. Prepare data models
The first step is to collect all the important questions that community managers would like to know the answers to and prepare data models for them. Engagement forecasting, user cohort analysis, community metrics benchmarks, etc. Anything that might be useful to a person who is working on a community, but requires knowledge of mathematics and programming and time to implement the idea.
2. Custom community analysis
The goal of the second step is to test and improve the prepared models on real communities by running data analysis on demand. Since the models themselves will be mostly ready, the main costs will fall on data preparation and presentation of results. This should reduce the overall cost and time of research at least by half.
3. Self-serve service
Anything that works for the on demand community analysis can be automated and presented as a self-serve service. Since people will not be involved in providing the service, the cost of such a solution should be minimal, making data science accessible to every community manager. Sounds tempting, doesn’t it?
In fact, I have already prepared the first demo version of such a self-serve service. You can find it at app.practical-cm.com. In another article I am sharing in more detail what you can find in the application today.
I need your help! Let’s work on ideas for the data model together!
Building a truly useful service is only possible if we work on it all together. I would like to ask everyone reading this post to reply to it or email me at (nicolas.chabanovsky@gmail.com) about any problem or task they have faced when working on a community that could be solved with data. Are you interested in metrics? Maybe predicting the number of posts? Identifying potential high-rank users earlier? Easy! Just email me about your data needs and I will try to build the best model that calculates what you need.
Everyone’s input is very important. I would be very grateful for your help and promise to share everything I collect in one of the next posts!