Getting to Simple – A True Messy Data Story
Business insights can come from even the simplest data visualizations. In this age where we receive more information than we know what to do with, seeing just the right amount of information is all it takes to make the right decisions.
Let me take you behind the scene of SCN’s of Gamification efforts, and invite you to put on a Community Manager’s hat. As a Community Manager, you’ve heard from multiple Topic Moderators that certain members are abusing SCN’s recognition system by gaining contribution pointsinappropriately.
Although, this was only anecdotal evidence, you trust and value the judgment of the SCN moderators, and decide to do something about this (after you’ve verified and dealt with the offending members). The first step is to. Was this the right decision, and what were the results?
After the change was implemented in March 2014, you gathered and received positive feedback from our Moderator community:
You also have data – lots of data, telling you which member did what, at what time, in which space – over 1 million rows of data each day, collected over many months!
The first thing to do with so much data is to prepare it in such a way that you can focus on your target metrics (number of likes to comments & replies, change over time):
- Out of over 200 different events/actions that are tracked, you create groups of only events related to “Comment Like” and “Comment Create.” (In addition to comments, ‘likes’ to replies were also included in this change – you decide to focus on Comments, or else the 100s of thousands of messages/replies will skew the scale of your analysis – trust me, a fellow community manager did the same and is passing this wisdom on to you.)
- Next, you create a time hierarchy so that you can aggregate the results by monthly totals, and select specific months for your analysis. This will allow you to further focus on your target metrics.
With a simple line chart to visualize the data, the impact of this change is clear.
- “Comment Like” events (green line) dropped by about 66%, and then stabilized.
- “Comment Create” events (blue line) remained stable over time.
This tells us that by make a slight change to the gamification mechanics, we were able to reduce unwanted behavior (artificially liking comments to gain points), while not impacting wanted behaviour (authentic creation of comments to other members’ content). Note the increase in “Comment Like” events in March when we made the announcement and stopped providing points for likes to comments. My guess is that some members were testing to see if we made the change properly. 😉
Visualizing our data in this simple way confirmed what we were seeing and hearing from moderators, and validated our decision. We then asked ourselves, “What if we designed missions that would channel the competitive energy of our members, and created awareness for some meaningful community topics?”
This led us to introduce missions with low-barriers, that allowed all members to participate regardless of whether they are beginners or topic experts. The Doctors Without Borders community challenge is a recent example of how our community is benefiting. We hope that making it easy to earn a badge will help increase awareness for this global program, and will result in benefiting Doctors Without Borders through your donations.
Thanks for your willingness to put on the Community Manager hat, and for reading this blog! If you’re attending SAP TechEd && d-code, be sure to join my session “Wrangling Messy Data,” where I will share more stories of how simple Lumira visualizations have given us insights that have translated to higher member participation, content quality and satisfaction on SCN.
Consider this my “House of Dragon” entry to the DataGeek III: Thrones of Data challenge. 🙂