I was in my office looking at some marketing data when my son walked in. “What are you doing?”
“I am trying to get some insights from this data” Being curious (and slightly bored), he asked “What do you mean by insights?”
I started saying “I am analysing this marketing data to find odd and interesting patterns…” and stopped myself. Instead of explaining contingency table and normal distribution to him, I asked him instead “Imagine you are tossing a coin. What is the chance of getting a head?” Yes, kids are very smart nowadays – they are already learning about probability in Year 3.
Now imagine we keep tossing the coins and we keep getting heads. At how many tosses should we question whether the coin is rigged? 2 times? 3 times? More? Data is just the information we collected about the coin tosses. Insight is the ability to tell whether we can think the coin is rigged or not.
Within an organisation, we have vast amounts of data about our customers at our fingertips and more are being collected each day. With so much data available, it can be difficult to separate useful insights from noisy data points. While many applications we use have reporting tools, it is often up to the users to build and interpret these reports. These applications are often standalone and as a result, the users may not be able to get the full picture. When we build a report to summarise the data, we often have a preconception as to what we are looking for and we may not see relating factors that are not obvious. Therefore it is critical for us to step back and try to understand what the data is or is not telling us.
Recently, we did some basic correlation analysis with the data from a client and in particular, we are looking for what factors have correlations with lead conversion. For example, we found there is no correlation between lead source and conversion rate, despite one lead source has generated significantly more leads and conversions. If lead source does not matter to lead conversion, maybe the business should focus on the most cost-effective way to market to these leads so the same spend will reach a wider audience?
So next time when you look at a report or dashboard, think about what you are really seeing? Are they just the data points or are there some actionable insights in there?