How Skillshare did three months of data cleanup in one week - Mixpanel
Product Foundations

How Skillshare did three months of data cleanup in one week

Jenny Booth

Skillshare is an online learning community for creators with classes covering forward-thinking topics in design, innovation, and entrepreneurship. They have millions of members that use their platform to learn cutting-edge skills, network with peers and discover new opportunities. That’s all great. Unfortunately, up until recently, they had a big problem with their Mixpanel data.

The problem: no one knew what the data meant

The team at Skillshare has been using Mixpanel for the past five years. But over time, as the people responsible for implementing and managing Mixpanel came and left the company, a key problem in Skillshare’s Mixpanel implementation emerged.

It’s a tale as old as time, and a problem that any customer using a Saas product to process, analyze, or act on large amounts of data has experienced at one point or another. In this case, it was a lack of clear documentation and guidelines for how to implement and name events and properties, that led to inconsistency among event names, and outdated data.

We spoke to Liz Yeomans, a Mobile Product Lead at Skillshare, who said, “An event would be named ‘watched class’ on web, ‘watched_class’ on iOS, and ‘watched a class’ on Android, making it really hard to remember what all the different events were.”

They decided to clean up their implementation, and immediately realized what a huge undertaking it would be.“We didn’t have a place to look to see what had already been implemented. We tried making lists and spreadsheets outside of Mixpanel, but the sheer volume of events made it extremely difficult and time-consuming. I didn’t know where to start,” Liz said.

The solution: easy tools to clean up the data, no developers required

Using Lexicon, Mixpanel’s data management tool, Liz and her team were able to start documenting and organizing their project, without bugging their engineers. They started by hiding outdated events, like unused events from versions that were three years old, directly in the Mixpanel UI. This made the event dropdowns cleaner, and eventually the only events there were ones that had been verified.

lexicon tags

Then Liz added event tags to categorize all the different events. She explains, “I went through and tagged everything and explained to the Product Managers that they didn’t need to look through all 300 Events. With Lexicon they could easily filter to the features they owned and quickly see what events we already had, and what new ones they might want to add”. Once they saw how easy it was to manage events for the areas of the product they owned, they were inspired to do it themselves, and became invested in building out Mixpanel to track their upcoming features, which helped the audit go by faster.

The ability to add definitions to events and properties also helped everyone – the marketing, support, and content teams, even new employees – understand what the data in Mixpanel actually meant. “The definitions help anyone at our company start using Mixpanel, with little training.”

Plus, they easily renamed events in seconds – without any developer resources using a more intuitive, consistent naming scheme, which functions as a reference for implementing new events going forward.

The outcome: time savings and more confident decision-making

“This audit and clean-up process would have taken us at least 3 months to complete.  But with Lexicon, it only took a week, and turned a tedious audit and re-implementation that requires engineering resources into a simple (and fun!) process that any PM or Data Analyst can do.”

Now everyone at Skillshare – from PMs and engineers to marketing execs and our leadership team – can check Lexicon to see what things are called, how often they are being logged, and if the activity is being analyzed. It’s the primary event-tracking tool that allows us to understand how our users are interacting with our platform – ensuring consistency in our event data helps us continue to improve our culture of data-driven decision making,” Liz said. Now that those teams can work from the same set of data, it makes it easier for the entire company to stay aligned on goals.

“Lexicon is built for real people and the real problems that companies have, like turnover and the well-intentioned engineer that implements an event incorrectly. That stuff always happens, it doesn’t matter where you work. All the mistakes we make, or the way data has a tendency to be good, but not perfect, Mixpanel has anticipated, and now I can just fix them with Lexicon.”

The main reason the team at Skillshare was able to use Lexicon so efficiently, is that every feature – whether it’s renaming an event, hiding an event, or adding a description to a property-  is easy for anyone to do in the Mixpanel UI, no coding required.

What’s next for Lexicon

Mixpanel is also excited to announce a few new data management features, available today. For enterprise customers, these include the ability to drop events and properties before they get ingested, so that they don’t clutter your implementation. Also, if you have three different events or properties that mean the same thing, like “Watch class”, “Class watched” and “Watched_class” you’ll be able to easily merge them into one. Lastly, all customers can easily bulk hide events and properties, like outdated or unused events, for a cleaner dataset.

Any Saas tool requires regular maintenance to stay up-to-date. If a new product ships, or something in the UI gets tweaked, or a feature gets deprecated – that likely calls for an update to your implementation. Unlike many legacy analytics tools, we’ve made it easy and painless to make those updates, so that you can ensure the data in Mixpanel is trusted, accurate, and usable.

Get the latest from Mixpanel
This field is required.