Gameplay Sessions: How to Design, Measure, and Analyse

Games offer the players an experience, just like movies or TV shows. In TV shows, you have episodes; a contained package of your story, in 30-40 minutes. What is the smallest bundle of experience in video games? A “vertical slice” as they call it in the game development cycles. The smallest, yes, but a complete representation of the experience you are offering to players?

It is one session.

One of your (hopefully) thousands of players, clicks the icon of your game, and the experience starts. Logos, then main menu; clicking “continue” or “new match” to warm up their muscles. After a couple of matches, they stop and explore their daily quests, claim what is finished, and then spend some time distributing their skill points. Some quit right here, others jump right back in to try out their new power, maybe passing a couple of bosses.

You get the idea. This one session actually has all the information you might need about all your game systems and how they interact with each other. However, most of the industry is only analysing the tip of this insight iceberg. First sessions as part of the “First Time User Experience (FTUE)”, or just observations about the frequency and duration of the sessions. Session analysis, when done properly, can give you so much more knowledge about your players’ behaviour. In this post, we are going to go over session analyses and try to uncover more of this area that is explored far less than it deserves.

First, I will give you a brief introduction to what I mean by “session”, and then I will walk you through a framework for session metrics. Once we calculate the actual numbers, we will discuss how to use and interpret those in two critical checks for your game. Finally, we will connect everything together by going through a pretty realistic scenario step-by-step.

What is a Session?

A session is the time from someone opening your game until they close your game. No matter the type of game you have, players interact with it in sessions. Some games have very short sessions (<10 minutes) while others can be a multiple day marathon (WoW World First races). Players interact with many aspects of your game, while completely ignoring others, in every single session they have.

You, as the developer, actually have all the control in designing the sessions and the experience the player has. You can control how fast someone progresses in the game, how much time they spend in the core loop, etc. Most developers think that players are going to play the game as they would themselves, but in reality the players are going to discover ways to play your game that you did not even think possible.

So even if you had to stop right here, give some thought about how you think your game is supposed to be played and how it is actually played. This can be done standalone and it would be insightful regardless; but we will of course connect these to actual numbers and metrics in the rest of the post. In true Data fashion.

What to track in a Session?

Now, I said that the sessions can and do encompass everything a player does. How are you supposed to track everything? How can we make sense out of that sheer volume of signals?

Good news is that you actually don’t need to track everything! But you have to be careful about what you are tracking and why. Broadly speaking, there are three categories of events you must track to analyse sessions (well, pretty much everything in your game):

• Core Gameplay Metrics (especially % time of Core gameplay in session)
• Progression Metrics
• Monetization Metrics You can have one of each, or multiple ones depending on what you are aiming for. There are a few examples below, but the critical thing is you have to have at least one of each. You can’t have a complete picture by missing 1/3rds of the player experience.

Your Opinion Matters… A Lot.

Once you have decided what your core metrics are, you actually do NOT touch the database yet. We got an invaluable exercise that frankly should be part of more analyses: You will guess the answers to the following metrics, to be compared with the actuals.

• Session duration & session frequency
• % of session in core gameplay loop
• expected progression in a session (e.g. 2 levels?)
• How many ads did they watch? Did they have any mtx purchases?

When doing this guessing exercise, you should try and have as detailed of an answer as possible and ideally, it should be in human-language and not analyst-speak. For example: “I expect a typical player to play our game in the morning commute and in the evening before bed, for half an hour each. Half of this time should be in the actual game levels, and they should be able to pass 3 levels in the campaign in each session. They will likely see ~5 ads and won’t purchase anything”.

This is not just a way to look back and admire if correct, or laugh if the numbers are vastly different. This is actually your expectation of how the game is played, and you can decide if you would like any of this to be different. “Actually 3 levels are too low, I would like them to progress faster so they can reach the endgame” is a totally valid design direction that is tough to discover without this specific framing.

Calculating & Interpreting Metrics

It is finally time to calculate the metrics. The numerical examples are going to make things clearer, so bear with me here on the theory.

There are many creative ways to use these metrics, and we are going to cover two of the really important ones: gameplay efficiency and pay-to-progress health.

Gameplay efficiency is how much progression your players achieve per minute of core gameplay. For example, a player plays 5 attempts of 3 minutes each in a 20 minute session. They pass 3 of those levels but fail twice. That means you pass one level every 5 minutes of gameplay. Whether this is a good ratio is up to you; maybe you want a slower progression through your game, and thus need to adjust your level rewards down.

Things get really interesting when you group players into low efficiency (bronze) and high efficiency (diamond) categories - what makes them differ? What is the 2 minutes per level group doing differently than the 5 minutes per level group? Is it based on player skill only or do we have any issues with difficulty curve? Gameplay efficiency, and various slices of it you may check can be incredibly useful for your game matchmaking, level design, and player segmentation.

When you apply a monetization perspective to the session metrics, you arrive at your games’ pay-to-progress health (P2P health). P2P Health is the progression boost a player gets by spending. To be frank, defining and categorizing players by “spending” is tricky with no correct answer. A good rule of thumb though, is whether they have spent in the last 3 or 7 days until this session. You don’t need the perfect classification, but you need a clean split of payers and non-payers that is defensible.

Once you have the classification of players, you check the session metrics among spenders vs. non-spenders. Imagine you have your non spenders complete one level in 5 minutes on average, while your spenders complete one level in 3 minutes. Spending increases your progression speed by 65%, and you have to keep this difference in mind when planning your new content releases. Take this same approach and break it down by the packages and items you are selling, and you get the beginnings of a full on game economy review!

These are just two examples of investigations that can be done; but you can get creative in however you use your session metrics! Want to have a “xp points per enemy defeated” as your efficiency metric? Go ahead! Just always make sure this metric is relevant to and representative of your vision and how your players engage with your game.

Scenario & Case Example

Now, as promised, let’s walk through an actual example. This is a hypothetical scenario that I have seen very similar versions of, and likely happened quite a few times across mobile game companies in the past couple of years.

For our example, we have a strategy game on mobile, that has been going strong for the past 3 years. Recently you noticed the growth of the game has slowed down, and realized it might be a great moment to have a look at the sessions to see if there is anything that might help!

As first step, you decide on the progression and core gameplay metrics, and write down what you expect their values to be. Then you calculate the actuals; some of them are matching what you had in mind but some are quite far from it. Here are the most divergent ones:

These already tell an interesting story. Your players are spending less time in the actual core gameplay, but they are progressing faster. They are also watching more ads than you thought they were.

Now this is not actionable in any way… yet. We can check these by your player progression segments (new, mid, elder), and notice this effect is much worse for new players while your elder players are mostly fine. Or. we can plot these over time and discover they started to change back when we added a new rewarded ad placement. Now we have a theory to test: the rewards from that new placement is too high for new players, and they don’t feel challenged and lose interest quicker.

This hypothesis is impossible to build by retention analysis alone; new players are still coming in and we may notice a drop, but can’t identify why. It is also not possible by the analysis of the new placement; everybody is constantly using it and it is generating a ton of revenue; we have no idea we are hurting players.

This is the advantage of having a deeper look into your sessions. You see how these things interact with each other at the most granular level.

Conclusion

Sessions are the smallest complete experience a player has with your game and the experience you want to provide to players. While we tend to have an idea about how to analyse it, we can and should go deeper into session data. In this post we have covered a straightforward framework to capture 3 areas (core loop, progression, and monetization) a great exercise in understanding your own expectation of your game vs. the players’ experience, and two important ways to read and interpret player sessions. The insights you gain from this holistic session view are not achievable by industry-standard approaches and that is what makes this so powerful.

In the next weeks, try to do the exercise where you think about your own game sessions at least and see if there is anything surprising. If you would like to go the next step, point your data team to this post or contact me. Every game is different, but we can find your game’s specific metrics and review your players’ sessions. Your design and development scope will expand, and your players will have more fun.