Data analytics projects are often undertaken to achieve a specific objective or find answers to a set of questions. While it is important to have objectives at the start — as they help to define the direction of analysis — serendipity also plays a role in unearthing insights. Businesses today collect huge volumes of data from multiple sources, often from large numbers of people and at many different points of interaction.
When these large volumes of data are analyzed — in addition to throwing light on the questions that have been asked — patterns, trends, correlations, and causes that were not previously known sometimes appear. These are the unexpected benefits of data analytics.
Let’s take a look at some examples:
Discovery of new user profiles
A consumer brand chose a popular social media platform to advertise on, with the objective of acquiring new users. Based on data about existing users, the campaign targeted people of similar demographics as well as hobbies and interests — a strategy called “lookalike” on some platforms. Once responses to the ads started to come in they began to focus on those profiles that culminated in a purchase. To their surprise, they discovered that the profile of buyers who responded most strongly to their offer and chose to purchase the product was very different from the one they had initially built, based on a smaller data set.
Further analytics helped to understand why their product was appreciated by this particular profile and in what ways they were engaging with it. Understanding the user profile based on this data, albeit an unexpected data analytics benefit, helped them to further refine their product to best fulfill user needs.
Discovering new buying triggers
A candy company was examining sales data to understand events that triggered purchase. For example, it was well known that sales were higher during weekends when people are more likely to consume candy. Data analytics was also revealing how sales were affected by festivals and weather conditions.
During this exercise, the company discovered that sales increased each time there was a hurricane warning in the area. The data seemed to suggest that hurricane warnings trigger the purchase of non-perishable food items such as candy. This was a data analytics benefit — an unexpected pattern as well as a correlation that helped them to better understand their consumers and plan inventory.
The insurance industry faces a constant challenge in identifying fraudulent claims while also ensuring that genuine claims are settled fast. An insurance company was using a variety of tools for this purpose — credit score checks, third party verification, criminal record lookup, etc. These were being analysed in order to predict customer profiles with a higher probability of fraudulent activity.
During this process, they made an unexpected discovery. The company had audio recordings of calls made by customers to their contact centers. An analysis of these audio recordings revealed fascinating commonalities among fraudulent calls.
Data analytics was able to pick up certain phrases as well as tone-of-voice that seemed to occur more frequently in fraudulent calls. These findings help them to prepare contact agents to identify potentially fraudulent behavior during phone conversations in order to mitigate risk.
By integrating the data across various sources and overlaying it with an analysis of telephone interactions, the company was able to predict and prevent fraudulent activities with much higher accuracy than before.
Identifying new factors that increase the risk of certain health conditions
A healthcare data research project was examining causative and correlational factors for a particular health condition. Health record data for a large number of patients was analyzed, with a view to better understanding the risk factors. When the analysis was expanded to include a larger number of personal details, the researchers made an unexpected finding.
They found that, within a state, the risk of developing this particular health condition was higher among patients who resided in specific zip codes. This led the research to delve further into lifestyles, family lives, living conditions, and diets — factors that could be linked to the locality where the patient resides.
Now the research team was able to identify certain living conditions that increased the risk of developing this health problem.
As we see from the above examples, there are myriad unexpected data analytics benefits to be found when research is conducted using the right techniques on sufficiently large data sets. Data scientists need to keep an open mind and structure their studies in a manner that makes these unexpected discoveries possible.