Let’s talk about how, why or why you shouldn’t set up analytics for your organization.
First, do you actually need analytics? I speak with a lot of companies that say they want to be “data driven,” which sounds great (and is). I also talk to a lot who say they want analytics. When I ask “why?” I get a mixed bag. Some companies are getting them in place because they see others doing it. Some realize the need to work off of data rather than guesses. Few start where they should, which is with key questions based on their operations and their specific team.
Why? You can have all the answers in the world, but if you don’t understand the questions, or can’t act on them, they’re a waste of time. So, my recommendation is to start with two basics:
1) What do we want to know?
2) How will we act on it?
Those sound pretty obvious, but I promise you, few think both all the way through. Why does this matter? Because if you don’t know what your central questions are, you’re going to get analytics that are window dressing--or even worse, a distraction.
What do we want to know?
Let’s take an example. Game Developer A is excited to get metrics set up, and they’re getting a dashboard with K-factor and retention. What questions are they answering and how will they act on the data? When I ask the developer what their key questions are, they don’t have anything to do with these two metrics. They say “I want to know why people are dropping off my game” and “My board wants to know how much we are likely to make this month.”
Well, neither metric answers those questions. In this case, K-factor answers the question “how much do players invite others?” and retention answers the question “how long do my players last?” What this company really needs is a technique for tracking dropoffs (funnels or metrics on their specific game mechanics) and a way to forecast revenue (ARPU x MAU is decent, predictives are better).
So, step one is the key exercise of “What truly are our main questions?” Then, and only then, should you go out and start looking at specific metrics or solutions. And if all you need is “how many players do I have?” you don’t need to go farther than Flurry, which is easy and free.
But maybe you want more. If so, here are some of the questions developers should consider addressing:
How much money have I made this (day/week/month/period)?
How much money am I likely to make in the coming (day/week/month/period)?
How stable is my player base?
What is my churn rate?
Which players are likely to leave in the next period?
How much are each of my players worth?
Which promotions are generating positive results?
Do my players generate value among each other (Social Value), or are they relatively isolated?
Which marketing sources are leading to conversions? What is the ROI on each?
What mechanics drive player churn?
What mechanics drive player conversion?
What are the effects of the changes I’ve made?
Of course there are many more, but they’re often variations of these. And, happily, any of them can be answered with the right tools.
How will we act on it?
Do you know what you will do with the answers? If these are simply for internal reports, benchmarks, or reports to your board, great, you’re all set. If these take some action, are there systems and people in place? In other words, are the metrics actionable within your specific organization?
This is a big deal, and involves both your culture and your org chart. One thing I’ve noticed time and again is that engineers are not at all like marketers. They are rarely in sync, and are sometimes contentious with each other. Most often they have different goals, different mindsets, and different questions to answer.
The engineers want to know “what’s working?” so they can fix it or learn how to do it better. For example, Is my level any good? Are players using what I made? Is there a mechanic I made that is leading to revenue or hurting it? The marketers find all of that interesting, but since they can rarely do anything about it, their questions are usually more like What can I do to increase spending? How do I stop churn? What promotions work? What acquisition sources were the most effective? and What is the forecast for the community?
Now, given results in the right metrics, each of these two groups will take very different actions. The engineers, given direct evidence of a level, quest, or NPC working well or not, will want to go tweak the code. The more enterprising ones will make a partial change via an AB test and watch the results as some players use the new code and play, churn and spend more or less after. So, the analytics must have a way to watch the data unfold over time. Typically a graph showing the use of the specific mechanic is the simplest approach.
The marketers want to reach players directly, so they need the analytics to feed easily into tools like email marketing systems (MailChimp, Exact Target, etc.) or push notification systems. And any table will need an exporting tool to allow at least exporting to a CSV file, but it’s easier if they allow custom selection of records and exporting directly to the API of the marketing tool. In sum, these CRM efforts should be customizable, directed to a specific subset of players, and trackable over time. It would be a wasted opportunity to simply run a promotion and not see specific results from it. So, the ability to segment metrics later based on who received which CRM intervention is crucial. Here’s an example from our dashboards, showing the baseline average and the specific averages of groups getting particular CRM events:
Identifying the Best Promotion Using Individual LTV and Social Value
This is a very actionable result. The marketer ran three promotions and has clear evidence that one did well (personal spending and social value both went up) while the other two backfired. The CEO can say “Hey, more like that one, please.”
The takeaway here is - metrics have to fit the organization and its questions. If there is no one to change the code, or send the email promotion out, it’s just window dressing. Or, if these really are the questions the organization needs to answer and there’s no one to act on them, it’s a clear signal to add that resource to the team. Many cash-strapped developers are loath to pour money into analytics because they seem like a place where costs happen and only pretty charts appear. But done correctly analytics are actionable, measurable and can quickly turn an expense into a profit center.