Thinking about daily active users (and the limitations of this metric)

Daily active users (DAU) – or even better daily active customers (DAC) – are key performance indicators that are widely used in the industry to assess the scale (and potential) of a live game. And rightly so. DAU or DAC provide insights into the number of users engaging with your game on a daily basis. When you are thinking about ways to scale your game, you are obviously thinking about the amount of revenue your game generates on a daily basis. Improving the monetization performance of your game (daily iap conversion, arpdac, LTV…) is a key aspect here. But in order for your game to scale over the long run – for revenues to grow and you being able to invest into UA to increase the number of potential revenue contributors (read: active users and customers) – you need more people playing your game regularly.

Regularly is the key word here. DAC/DAU only tells you who is active on a given day. The limitation of this metric – like any snapshot metric (such as retention) – is that is doesn’t provide an adequate sense of permanence or continuity. By that I mean that you know who is active on a given day, but it doesn’t always allow you to answer a more crucial question: how many regulars are in your game? Are you building a “sticky” userbase on which you can rely on to anticipate future revenue? A portion of your daily active users are regulars – but there isn’t a linear relation between both concepts.

Don’t get me wrong. DAU/DAC is a crucial KPI to assess the scale of your game and evalute changes in its performance. But these KPIs are not perfect and they don’t provide immediate insights into who those active users are over time (remember, DAU is a snapshot at a given moment in time): Are those active users committed to your game? What is their level of engagement? To what extent is your game a part of your users’ day to day routine?

Ultimately, that last question is the most important one for you: to what extent is your game a part of your users’ daily routine? You rely on various KPIs to cover some of the blind spots and infer an answer to that question. But it’s also a very difficult question to answer. First, there isn’t an objective definition of what makes someone a regular and what doesn’t – that requires a judgment call on your part based on what is relevant for you and your game. Second, a DAU/DAC metric is very straightforward and unambiguous. But there is a discrepancy between who is active on a given day and who is active regularly.

The reality is, when you look at things, it’s most likely only a small portion of your active users on a daily basis are very committed and engaged. And here I’m discussing daily actives in general – this applies both to daily active nonpayers and daily active customers. To best illustrate the fact that DAU/DAC in itself is not always enough to convey a sense of your userbase – and who they are day-in day-out – you can look at your monthly or weekly uniques. More specifically, for every weekly/monthly unique player, how many days in a period does s/he play?

weeklyuniquesmonthlyuniques

 

To be consistent, you should only look at players who have installed long enough ago to have at least one entire period – players who installed 1 or more month (or week) before. You want to exclude your new installs because 1) they couldn’t possibly be playing every day of the period (if I install on August 27, I can play at most 5 of the 31 days in the month) and 2) you might be looking at 50% of your installs who will never play your game after the day they first launched it. Those players are not representative of your active users because they don’t find your game appealing enough to continue engaging with it after the first impression. That’s of course a problem, but it’s a different kind of problem that needs to be addressed in a different way.

There are a few general conclusions you should be able to observe regardless of the game (as I’ve said in the past, I’m discussing “mainstream” games. I have no experience and little visibility into the engagement patterns of hypercasual games – and I would imagine the engagement patterns in those games has some key differences compared to what happens in mainstream games). First, customers will have a higher level of engagement than nonpayers (not surprising). Over a given period, a lower proportion of customers will only play one day and a higher proportion of customers will play every day in the period. Second, you might be looking at an important proportion of monthly/weekly uniques who only play once in a given period. Third, players playing every day in a period are a different breed from everybody else. The longer the time period considered, the lower the proportion of players playing every day of the period. But that’s simply because there are fewer opportunities to “drop off” playing daily in a 7 day period than in a 30 day period. So, if doesn’t mean you have more flaky users when looking at things over a month rather than over a week. It just means there are more chances life will get in the way at some point over the duration of an entire month (compared to the duration of an entire week).

You ultimately are mostly concerned with the percent of users playing every day in a period. Incidentally, this can be a robust predictor of “churn” (for lack of better word). And looking at days played during the month – or more specifically, the percent of players that play every day of the month – can be a more detailed and relevant alternative to the DAU/MAU ratio. The DAU/MAU ratio is a way to assess the “stickiness” of your active users during the month (or whatever timeframe is your denominator). Those ratios vary between 1 (during a given period, every player plays every day of the period) and 1/total days in the period (during a given period, every player plays only one day). The closer to 1, the more your userbase is composed by a group of “regulars”. In other words, if you have a DAU/MAU ratio of 20%, that mean a monthly active user plays on average 20% of days in the month.

But looking at average times (both the time that has elapsed or occurrences of something) is misleading because it doesn’t reflect user behavior – it reflects how the accumulation of times averages out over your entire userbase. And looking at averages is not very relevant because you don’t bill by the hour or days played. Looking at how many days your active users are playing is important because when you start to follow the money, you should observe that most of the revenue in a given period comes from your most engaged players. That’s why the whole point behind this post is that looking at DAC/DAU alone doesn’t immediately convey a sense of how many very engaged active users you have in your game. Because what matters the most is to have engaged players over the long run, not just a given number of active users on a given day.

revenuesperdaysplayed
In the above graph the vast mojority of revenue comes from players who have played 31 days in the month. Players who played only one day during the month contributed the least (despite the fact that most monthly unique players played one day during the month)

 

And this also touches upon the engagement vs. monetization debate that permeates the industry. On one hand, if a user is playing 31 days during a month, s/he will have many more potential days to contribute to the title’s monthly revenue compared to someone only playing one day during the month. But when you look at relative monetization performance you should observe that players who engage the most with your game also engage more with your game on a monetary level relatively speaking. In other words, the biggest revenue contributors play the most. But – and this is key – they also spend more every time they are engaging with your game. The active day of an engaged player is simply worth more. You can look at percent of active days where an IAP transaction occurs, at transactions per active day or revenue per active day. Players who engage the most will score the highest on all those levels (and you’re looking pretty much at the same curve in most of those cases).

acgdailyIAP

One comment

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s