GAME Product Analytics Framework

Success metrics provide a quantified measure of a product is doing; They are a lens into the health of your product and its performance. Defining these metrics is therefore intrinsic to the development process. The challenge, however, is coming up with a metrics system from scratch.

Most Product Managers start from the bottom – collecting and monitoring a number of data points they can get. Then they move on from there, one step at a time. But just because you can track everything, doesn’t mean that you should. Some of the data is more valuable than the other, and truthfully speaking, there is a lot of insufficient data that can end up clouding your judgment.

The GAME Framework

This is a four-step process that can help you define the metrics for any product or feature. Let’s look at these steps;

Step 1 – Goals

This is an almost obvious step. To your metrics back to the strategic outcome you’re looking to achieve, you really need to start with goals.  For example, is your goal to increase the conversion rate for your service?  If so, declaring that is the first step. Placing the goal-setting as the first step ensures that you do a top-down analysis which is critical in metrics development.

Step 2 – Actions

Next, what are the user actions that you want to affect, as it relates to the goal? It would help if you defined the user actions or all the actions you would want the users to take with the product. This usually starts with a qualitative list.  But you should be able to describe behavioral steps that you wish to change.  For example, if only 5% of users proceed from free accounts to paid subscribers, that action of proceeding through subscription may be what you want to target. 

Step 3 – Metrics

In this step, you are now ready to turn the desired user actions into trackable and measurable value. Please note that you can bring in the data and engineering teams to vet your metrics if you haven’t already done this. They will provide you with technical guidance that is feasible. Here are some decisions for you to make with each action:

• Direct vs. Proxy – Ask yourself whether the action can be tracked directly or if you can use a proxy to measure the action.
• Individual vs. Aggregate – ask yourself whether you can group many different actions for a specific overview and then separate the slices for later analysis.
• Magnitude vs. Ratio – ask yourself whether it matters more if you measure the volumes over the actions.

Step 4 – Evaluating (the Metrics)

The best way of ensuring that the metrics provide you with the right insights into the product is to test and iterate. It’s hard to know-how metrics will behave unless you start collecting the data. One of the most critical evaluations you shall make is the functional usefulness of the metrics you have created. It is easy to check this by monitoring the metric. This can be done with false negatives and false positives.

For example, if your metric drops, there should be a signal to show a real problem in the product. Additionally, if an issue surfaces, the metric should flag this issue. Different behaviors in the metric will be used to inform the correct iterations in the GAME steps.

Conclusion

These days, there is a “Move fast and Break Things” way of doing things, and it can be tempting to only concern yourself with creating the metrics after the product has already been launched. However, metrics should be used to define the process before you even start development, and what you find should translate into huge benefits. This is because:

• Metrics will help you maintain the proper focus.
• Metrics will help set a common goal for the whole team, which they can rally around.
• Metrics help to hold the Product Manager accountable to the business and the users.
• They enable the team to have more objective-oriented decision-making in the launch period of the product. This includes prioritization and triage.

Deciding to fly blind during any product life cycle stage will most certainly decrease the team’s ability to learn. In fact, you cannot improve what you haven’t tested first. It is good to have the correct data upfront as it will make the difference between the failure and success of your product.