League of LegendsMobalytics

Lessons Learned from 2018: A Mobalytics Retrospective

It’s been very hard to find the time to write about what we are doing when we have so much to do, but I have come to realize that this is a backward way of thinking. There is always much to do, and sometimes taking a step back to write is the best thing to do it allows you to move farther away from the problems and see them in a new light.

It’s like opening your map when you are stuck in a dungeon. Imagine running around the water temple in Zelda without checking the map…hindsight is 20/20, but let’s talk about 2018 coming into 2019.

Premium has launched!

We launched our first premium feature, Progress Analysis, before the holidays, and although we didn’t really take a holiday, all of us had time to step back spend time with family and reflect.

First, I want to say thank you to all our supporters who have been there since day one and for those who chose to upgrade to a subscription, we are humbled that you feel like Mobalytics is worth your hard-earned money.

It was very fulfilling to deliver on a promise that started over two years ago. However, this promise doesn’t end here. We promise to continue to make the Mobalytics premium experience the strongest collection of features to help you achieve in League of Legends.

Next on the horizon for premium subscribers is our Lesson Plan feature which we have been conceptualizing for more than 6 months. I can finally say that we have settled on what will be the first shippable version.

However, our goal is not to just add new features, although we will be planning more after Lesson Plans. We are also committed to taking existing features (free and premium) and making them better. We listen to feedback closely and have improvements already in the works. Our community members have been integral in allowing us to push boundaries and get to a polished and multifunctional product.

Today, I would like to share with you some of the lessons we learned over the last year. I hope that you can enjoy our struggles just as much as we did, and learn from them. I can’t possibly tell you everything so I will focus on a few stories.

Lesson 1: Research and analytics are cool but simple things are cooler

This was a hard lesson to learn. Our first version of the GPI was the culmination of hundreds of hours researching metrics, analyzing data, creating and tweaking formulas, and making sure that what we deployed FELT like an accurate analysis of player skill to even the most basic player.

I’d say we succeeded in that respect, but people couldn’t really just use it or understand the details behind the numbers as clearly as we would have liked.

Let me give you two examples that fall on opposite ends of the spectrum:

Example 1: What’s a good score?

The less analytical player would come to us and ask: what is a good score? Although we included a graph showing player skill and we used a rating system from 0–100% (that you may expect most people are familiar with), a lot of people just couldn’t put their finger on EXACTLY what the score meant.What's a good GPIAm I as good as a Plat player if my score is the same? Am I Diamond level if my score is higher? How is that possible… when I am in Bronze? What does a 55 mean?

Explaining that our scores are based off the average from a huge percentile database and you can be above average in one category and there are many reasons why you can be good at one skill and still lose is HARD WORK and not something I want to make a day job out of.

Example 2: Give me more!

More analytical players are more curious when it comes to data. Where does this number come from? What are the weights in this formula? What if I die but my teammate was nearby, does it count as a pick? What if I am a split pusher, I don’t need to have this skill so why are you showing it to me? You get the idea.Formulas questionLeague is a very complex game with an infinite labyrinth of situational dependencies, and contextualizing those dependencies is what we do to create our scores. It’s natural to expect to start from the more basic situations and layer on increasing edge cases, but scores, advice, and insights need to be specific enough or they become worthless.

How to keep everyone happy?

The GPI has gone through many changes since it’s launch and has another big update on the horizon. We will continue to see many iterations as we learn more and get more data.

The new changes in Mobalytics 2.0 allow people to interact more easily with the GPI and advice to understand where their scores are coming from and why they are being told to focus on specific things. The GPI will always be the engine of our analytics platform but that doesn’t mean that the engine needs to be what you interact the most with. Automotive brands with great engines definitely let you know what’s under the hood, but most people don’t really care: they want their shit to work and get them where they need to go.

The challenge is making sure that gearheads can tune their car and bask in the glory of the details, while the regular joe can drive their super cool car to work. We applied a similar philosophy to move forward. You will notice that the new homepage is your Dashboard. This is a first step in moving the GPI under the hood.

In the future, expect to see more easy to use applications that are powered by the GPI, rather than you having to go to the GPI. The GPI will always be your digital fingerprint that you can show your friends and check out other people with though. 🙂

Lesson 2: Kill your babies

If you spend time reading about game design or tech product development, you will already be familiar with this. It takes a lot longer than you think to make something people can use seamlessly. In our case, we have so many good ideas that need to be done. We get constant affirmation of this through our user requests on things we already are thinking about, but it’s easy to let that affirmation tunnel you to execute without thinking critically.

In the last year, we have had to make so many hard decisions between features we consider essential, not to mention features that would be great. When you are looking at two features that you deemed absolutely necessary and you realize you can only make one and fine tune that one, decisions become hard and everyone is going to weigh in. Here are a few of those decisions and what we were thinking when we made them:

Post Game: Abducted without a trace

Post Game is one of those features we get asked about a lot. In case you forgot here is what our Post Game looked like:Retrospective Post GameThere were a few things we just were not happy with. Premier among them was the “Reflection” advice we were giving you. The advice was just not specific enough to be meaningful, and without enough branching pathways in our algorithm, the advice was prone to repeat itself way too often. Without a meaningful insight, our Post Game was just a different looking match details page, so right now we decided to redirect people to a match details page until we can bring back the post game.

The second thing that really bothered us more than most users (although some users were perceptive enough to notice), was that we were measuring our scores as functions of multi-game aggregates. So, it was easy to score someone based on their performance in 5 or more games, but single game scores were either inflated or deflated (which lead to a lot of 100’s and 0’s). For example, it’s challenging for a person to average 1 death or less in 20 games, but it’s’ rather common to play a game where you get 0 deaths.

Why didn’t we just fix these things and keep the Post Game? At first, that’s what we thought, but when we tested the waters, we realized that we would need a lot more time than anticipated. With limited resources and other great features we wanted to work on, we made the decision to pull a feature we were not satisfied with. When the Post Game does return (SOON™), it will really match the vision we originally wanted and be able to generate meaningful insights based on what happened in that specific game.

What we learned

We have no shortage of great ideas, concepts, and prototypes here at Mobalytics. The team is constantly riffing new shit that deserves to see the light of day. The problem is that simple prototypes don’t cut it as final products. When you, the users, finally use it, there needs to be polish and many moving parts that are carefully thought through. How does it integrate with our databases? How do we handle prioritization? How does the player interact with each part of the UX? How do we make it responsive and smooth?

Answering all these questions takes time and energy. We have to decide which babies are more deserving of our energy, and although it’s not fun, it is necessary to make sure you guys see polished and quality products in a timely way.

Lesson 3: Make sure to do the basics

What are the basics? In our case, it’s not overlooking the things that the player-base takes for granted. It’s also making sure that we have the proper routines in place to keep up with a constantly evolving game. We have made a lot of mistakes here since we started, but we learn quickly.

Builds are the building blocks we forgot

When we first set about to design our feature set, our logic looked like this: “There are all these sites and they tell you champion stats, your match history, and popular builds. We need to make sure that we do stuff that people need but can’t find anywhere else.” That missed the mark…

Although people definitely dug what we were doing, we sorely underestimated the value of having a single repository you can easily access the things you need. One of those things is builds.

League is a complex game, with 143 champs, each with several different builds that change as the game changes. People are constantly leaving, coming back to the game, and trying new champions. Being able to easily find a build and use it with no hassle is too important to not have for players looking to level up their League experience.

What are we doing about it?

We scrapped something together and added builds as a hasty plug-in to the Pre Game. This definitely did not do it justice, and one of the first things we are planning to do is deliver meaningful build info where it most impactful. We have some cool stuff planned to take builds to a new level not yet explored by what’s out there. We are also making a commitment to prioritize fixes and updates on the platform as the game changes.

Final Words

Season 9 is upon us, and things drastically change with the onset of a new season. This time we will be ready for it, but your feedback is essential to our continued improvement

Well, I hope you got something out of this. I really enjoy when people openly share their struggles and we can come together to discuss how to be better. That’s always been the impetus for our platform and the community we built around it. You can count on us continuing to improve, just like you, for the year to come.


Uthgar, Mobalytics Warchief of Science