Flavors of Analytics
In an earlier article (Flavors of Prototypes) I heard from many people about how it helped them to see prototypes in a new and more powerful light. Even though I had written previously about each separately, putting them together can help people get the big picture.
So in this article I thought I would do the same thing by highlighting the main flavors of analytics. When it comes to data, I find many people have the same behavior. They think of just one type of analytics (usually web analytics) and can miss the bigger picture and larger benefit.
There are of course many different types of analytics, but here are the four main types of product analytics that I consider critical for product work:
1. User Analytics
These analytics help us to understand how our service or mobile app is actually being used by real people. We usually look at this data in aggregate, but we can also view individual journeys. Maybe you’re trying to understand if your on-line tutorials are being used. Or maybe you are considering removing a feature and you want to see how frequently it is accessed. Or maybe you’re tracking progress through your funnel.
Most teams are using either Google or Adobe tools for this, but there are many other options as well.
2. Customer Analytics
For most consumer products, the user is the customer, but for most business products, there can be many users and in fact many types of users.
Customer analytics help us understand how our product is used across the customer’s business or enterprise. For example, how many users are active at each customer? Or, how many support calls are placed on a per-customer basis? Or, how long is it taking a customer to get up and running with our products? Or, what is our average revenue per customer?
Some of this data can be pulled from the user analytics tools, but often we get customer analytics from our CRM tool, such as Salesforce.com. We also can get financial information about our customers from our financial systems. It really depends on what tools different parts of your company are using when they touch the customer.
3. Business Analytics
Most analytics are snapshots in time, such as what happened yesterday. However, it can be extremely powerful to look at how our products are used across time. Such as the lifetime value of a customer, or the customer churn rate, or the customer acquisition costs across all sources.
Most companies aggregate the data from their many key sources into some form of data warehouse. The user analytics data, the customer analytics from their CRM system, their financial data from their financial systems, and so on into a single source that can be used to look across data sources and across time.
If your company has data scientists this is probably where they spend most of their time. There are lots of tools for managing and accessing the data warehouse, some old and many new. It is very possible that your keys to success lie in your data warehouse.
4. Compute Analytics
There is another very different type of analytic that I find hugely insightful. Suppose you have a theory about how your product might be used. For example, maybe you suspect that the customers with the highest lifetime value have the most information stored in your system. Maybe today you don’t have a way to report data usage by customer. So maybe you instrument your engine to be able to report this data usage and then you can try to correlate.
Or a simpler and very common example has to do with performance. Maybe you know certain operations are slow overall, but you don’t know which parts are the real issue. So you instrument and track the components.
By its nature, this work is usually completely custom – the instrumentation and the reporting. But you don’t want to limit your analytics to just the customer touch points.
These four are by no means the only types of analytics. You can also track process analytics (how is your team doing – things like velocity and quality), or sales analytics, or many other forms of financial analytics. But these are the four main types of product analytics that I find value in for most teams.
The product leader does not have to be the company expert in each of these areas, however, he should be capable (trained and granted access) to use each of these. Moreover, the product leader absolutely should be the primary person that understands overall the data about his particular product. While others in the company are there to help with this (especially user researchers/business intelligence/data analyst/data scientists), the ultimate responsibility for this is not something to delegate.