While sophisticated statistical methods for data-analysis are the current focus of many business discussions, most practicing managers neither have the expertise for such analysis nor the time to learn the methods. And while their data-handing capacities remain limited, the data itself are growing by leaps and bounds. It is tough to handle the pressures of day to day operations as well as make highly informed, data-driven decisions.
Based on my experience in the data-intensive Telecom industry, here are some principles on data analysis which can help you get more insights with no extra training.
1. Keep an Open Mind
This means doubting – ‘good sense’ – which is simply a habit of thinking in a given context. Make it a practice to doubt everything you hear about the business – even from veterans. Articulate the common assumptions and subject them to scrutiny of hard data. Sometimes, they may hold true. But when they don’t, you will end up with useful insights.
2. Predictions First
Before looking at the numbers, make predictions based on your current understanding. Create as many hypotheses as possible. This makes you think rigorously about your ‘world-view’ and uncovers your assumptions about how the business parameters are related to each other. Testing these against the data will will help in sharpening your intuition in the given business environment.
3. Benchmark in the Beginning
Once confronted with a data set, begin by identifying simple benchmarks and descriptive values. Namely, minimum, maximum, and the average. Listing them out separately will help in creating a general ‘sense’ of the data, and help in identifying outliers – which are another source of business insight.
4. Compare, Compare, Compare
The division operator(/) is the magic wand for revealing insights – it essentially means comparing different fields. List out all metrics for your business. Some of these will be primitive – like ‘Total Customers’. Others will be ratios – like ‘Number of Customers per Month’. Your task is to come up with new ratios – compare every parameter with every other parameter. For example, ‘Sales per Salesman’ or ‘Customers per Region’. This can be a huge time consuming exercise – and it is also the most productive.
5. Beware of Averages and Summaries
While they are important to begin with, do not be satisfied with just averages. It is crucial to look at the distribution. Remember the 80-20 rule? Check out how does it apply to your business. Maybe it’s seventy-thirty, or ninety-ten. This helps in identifying the areas you need to focus on, as well as other business trends which are affecting the distribution.
6. Watch out for Spikes – especially the Good Ones
A good principle to follow here is – “If it is too good to be true, it is not true”. Our general tendency makes us feel good when we see great numbers on a parameter, and thus we ignore the details. However, it helps to look under the hood here too.
7. Data and Human Behavior
Beware of the idea that a huge amount of data can tell you all the stories. If you have no context, it will be more than useless – the interpretation will be error-prone. Talk to the people who have been reading the same data for some time to understand the hidden hints, quiz them on the outliers to deepen your comprehension. This will make you better at understanding the behavioral traits underpinning the numbers.
After all, business is more about people, and less about data.
What techniques have you discovered to learn more from your data? How do you reach uncommon insights? Looking forward to your comments.