Data analytics has become pretty much pop culture today; everyone is talking about it. But not all seem to practice it the right way. And that’s when we come across mistakes that are so common in this field.
So, here are a few tips to recognize these errors and work around them in time:
Scouting for answers in the wrong places
Most people in Business Intelligence focus on the scale of data and tend to track metrics, even before they thoroughly understand the business. Simply put, it’s like baking a pie without acknowledging the occasion, but with the right cups and tools. So here’s a tip! Get familiar with the business – it’s core competencies, weakness, and the industry standards before you do anything else. Next step- explore all the metrics, it’s relations and if anything’s amiss within the existing reporting system. Once all this is done, it’s easier to swim through the data and deliver useful insights.
Correlation is not the cause
Just because two metrics are correlated doesn’t mean one is the cause of the other. One tends to believe that if the bottom of my data funnel has dropped, then the reason behind this drop is a decline at the top of the funnel. But to find an explanation for outliers, there needs to be a cause and an effect. The cause here, is your reason, whereas the correlated metrics are the effects.
Tracking too many metrics
Keeping a track of too many metrics and diving too deep into data may not always be fruitful. You may get lost in a massive pool of data and not fetch the kind of results you desire. More often than not, the right answer is right in front of you, but only if you track the right metrics. Keep an eye on important, comprehensive metrics and everything else will fall in place.
Using insufficient information to extrapolate
Predicting performance is an essential part of any business analyst’s job. Having sufficient amount of good quality data, and concrete assumptions are vital for predictions. But beware- using insufficient data for extrapolation can be dangerous. To cite an example: You can’t assume that if a family wins 1 lottery on the first day of the year, they will win 2 lotteries on the second day of the year. A good example, on the other hand, would be using the data for gold carat and price to predict how much X gms of gold would cost.
Relying only on quantitative information
Ignoring an event of the past or any upcoming event, along with their effect on company metrics, will almost always backfire. Current metrics that might look good because of a seasonal event of the past, need not necessarily mean that future metric trends would remain the same. These events always have to be accounted for while setting targets or predicting growth. For example: If we have a Diwali sale in October and no sale in November, then it’s not right to expect a month-on-month growth from October to November. Ideally, we must attribute some proportion of the performance to the Diwali sale and then take X% of growth from that for November.
Ignoring new and more powerful technologies
Getting too comfortable with the tools you use is convenient. But if you continue this practice, your knowledge might get outdated, just like the technologies you use. Say for example Most analysts tend to start with MS Excel for majority of their analysis. As the raw data starts getting bigger, excel spreadsheets start crashing (which obviously isn’t positive). So, they tend to move on to R or Python, basically more powerful softwares. Our tip- always keep exploring and pushing the limits!
Assumptions are an integral point of consideration when formulating business strategy, as they can make or break your project. So, as a practice, always compare predictions with real outcomes and critically assess how predictions could have been better.
Hoping that these tips help you avoid common mistakes and make you a smarter Business Analyst!
-Authored by Sahil Verma, Senior Business Analyst at Haptik.