Trying to find the cause of a problem in your business? Data can help, but sometimes knowing how to explore and interpret it can be intimidating. We've put together this no nonsense data analysis guide to walk you through a simple process so you can confidently use data to find answers and make smart decisions.
In this guide: Houston, we have a problem. You’re in a hurry. You need answers, NOW. As the head of customer support, you notice a significant increase in ticket response time and you need to know what’s causing it.
Perhaps you’re the marketing manager for a SaaS company and see that signup numbers have dipped. The CEO and VP of product want to know the cause. Maybe your first task of the day is figuring out why the average cart abandonment rate for your ecommerce business is increasing.
Or suppose the activation rate for your mobile app has decreased and you’re responsible for figuring out why. Whatever the problem, figuring out what caused it and how to fix it is now your top priority. You already know data can help solve the problem, but you don’t have the time or expertise for a massively complex data investigation. Good news! You don’t have to be a statistician or have unlimited time to solve your most pressing business problems using data. This no nonsense data analysis guide will help you confidently draw conclusions and make smart, data-backed decisions. Leaders need to focus on intelligently sifting through the massive amounts of available information to retrieve knowledge that is actionable, and to use effective processes and tools to make smart decisions.
Ronald van Loon , data scientist, speaker, author, and founder What’s going on? No, really. Before diving into any kind of data analysis, you should quickly validate the problem you’ve identified. The single most critical principal I apply when analyzing data is a rule my high school math professor taught me at age 14: ‘Don’t write the first line of code until you can describe in plain English the problem you are attempting to solve!’ Simply put, if you can’t explain in plain english the business problem you are setting out to address, no amount of data analytics is ever going to solve it.
Dez Blanchfield , investor and data scientist. Could this issue be a symptom of a bigger problem? For example, is the dip in signup numbers an indication of a website glitch? Could the increase in ticket response rate actually be an indication of a deeper staffing problem? Look back over a wider period of time. Is this really an outlier? Answering these questions is particularly important if someone else has reported the problem.
On the flip side, is this issue a freak instance such as a reporting error (i.e. selecting the wrong date or a bug in the reporting software)? Have other related metrics dropped off similarly? If you notice downloads have fallen off a cliff but activations haven’t, perhaps downloads aren’t being captured properly? If a metric is counted in multiple systems (e.g. Google Analytics and your own event tracking), do both systems show the same drop off? Make sure you’re looking at a metric that matters. Rates are a great example of this. You might notice the website conversion rate has dropped but if the raw number of signups hasn’t fallen, then this goes from an emergency to a mystery to uncover!
This quick, preliminary assessment answers two questions: is this actually a problem? And if yes, what’s the core problem here? Think of this as the data analysis version of ‘a quick web search’ to confirm that yes, this is a problem worth looking into further.
Why might this be happening? Now that you’ve verified the problem, it’s time to tackle the cause.
1. Look for quick wins. Similar to the standard tech advice to ‘turn your device off and back on,’ look for any obvious possible causes or answers to the problem.