Analyse This: An expert guide to making the most of game analytics

Analyse This: An expert guide to making the most of game analytics

By Dr. Maria Sifnioti, Total Eclips

August 9th 2016 at 11:15AM

Dr. Maria Sifnioti, producer and researcher at A Clockwork Brain 2.0 developer Total Eclipse Games, offers her advice on fully understanding analytics and using collected information to refine your title

Analytics is a truly invaluable tool. With proper use, it will enable you to make your product even better for your audience. This will lead to more effective monetisation.

The plethora of game analytics tools that have become available in the last years, and the lower barrier to entry in terms of both pricing and training, really leave no excuse for not integrating one in your game.

Like another crucial tool, usability testing, it can assist in discovering issues, and pinpoint problem areas that the creator would not have been able to identify on their own. With that said, it is not a panacea and can offer guidance on certain questions, but not others.

After initial integration, most analytics tools, are set to automatically calculate the usual metrics (sessions/D1/D7/D30/ARPU/ARPPU/MAU/DAU). But what about the metrics that you need to code into your game? I will discuss some key game areas, provide a set of starter metrics and questions and, where applicable, provide examples.

First Time User Experience (FTUE)

What do new players do in your game, within the first five to ten minutes? If you only have resources to track one area, this should absolutely be it, particularly if you have a free-to-play title. It does not matter if you’ve polished your selling points, because if your players are leaving, they will never get to buy anything.

Your main aim here is to monitor dropout rates at different stages in your game. Apart from the obvious suspects (tracking taps to play, continue, restart levels…) look out for ‘hidden’ influencers such as loading times, possible lag issues, time a player remains in the level and so on.

If you are running a multi-platform game it is often worthy to compare analytics by platform (or even device type such as tablet/phone) as you might get significantly different dropout rates, so make sure that you track core device/platform stats as well. Comparing your dropout rates with reports from a crash-tracking tool is also a good idea, in order to gauge if you might have a bug or crash that may be causing trouble.

"What do new players do in your game, within the first five to ten minutes? If you only have resources to track one area, this should absolutely be it."

Dr. Maria Sifnioti, Total Eclipse

Data-mine your returning players

After your FTUE analysis, you know that a number of your players are leaving. But what about those who are staying beyond D1 and D7? What do they do in the game? What have they discovered that your early players have not? Can you bring that feeling to those first few minutes?

When we analysed the behaviour of our own returning players, we discovered that many of them were playing a particular mini-game. Originally, the game was towards the end of the selection list, so players had to go through all the other games and pick it out; so we brought the game to the front.

Player Progress and Accomplishment

If you are developing a level-based game, what is the ratio of starts/completions/losses for each level? If you have a more open game, think quests, missions, anything that drives the player activity forward.

By tracking success and failure rates for all your levels (not just in FTUE) you can gain insight on the player's’ sense of progress. Do they get stuck? Is something too easy and they get bored? Do they leave things unfinished? Metric complexity can also become more elaborate if you have consumable usage (boosters) that can tilt the balance.

If, for example, your players were also progressing with levelling up their character, and getting rewards for doing so, you’d need to track more metrics: the time it takes between level-ups, possible actions that they do in-between, and of course, identify possible drop-off points.

Measuring Earning and Spending

Players may earn consumable resources and then act to spend them in the game. Some might also spend real money (if you have a free-to-play title) to get bundles of resources or other items.

Whatever economic model you have chosen, you can use analytics to measure it and improve it. Some starting questions you could ask:

Where: Do they find a way to access them? What do they do while there? Track all UI areas where the players can collect and spend resources

What: What is being bought more often? Used more often? Ignored? Are there differences per player experience? What is the time spent in-game?

When: At those key events in the game where you prompt players to spend real money, do they go for it? If you have offers in place, do players take advantage of them? What about time-based offers?

How much: Are players spending real money? What do they buy? Do you monetise non-spenders (for example, with incentivised video ads)? Are your ad placements getting a good conversion?

Who are your buyers: What did they do in-game before they made a purchase? Have they explored different features than non-buyers? How much time have they spent in the game before buying? Are there any repeat purchasers?

Conclusion

Never, ever, release a new analytics integration to a live audience before testing it first. Have a thorough QA process for every new, or updated, metric.

It’s really bad to discover that you’ve made a typo and you’re not tracking any data. Even worse, that you’re tracking the wrong data, which could lead to erroneous decisions. At the very end, the quality of your data will determine if you can trust it.

The use of analytics is a loop between knowledge and action: collecting data has no point if you don’t evaluate it and decide if you need to take action. Conversely, decisions can, and should, be supported by data.