There are multiple reasons why a big data projects fail:
Analysis Failures
Not asking any questions pertaining to the data
Back in 2008, Google started predicting the trend of flu, and declared the outbreak of an epidemic, weeks before the CDC. Again, a few years later, Google reported an over-approximated doctor visits report, which was inflated by at least 50%. Internet users, rather than questioning the data displayed by Google, followed the trend started by the Internet Search Engine giant.
Not asking a single right question
Once, a car manufacturer, that owned several car dealerships all around the world, set out on a course to a project of sentiment analysis. The projects cost amounted to USD 10 million, and the time taken was around six months. Once the project yielded results, all dealerships were contacted and asked if the results do have any effect on the total annual sales. Sadly, the result was proven to be a wrong one, as no such effect existed.
Making use of wrong and incorrect models
A banks PhD decided to turn to other sectors for the purpose of searching for big data related successes, and incorporate such successful ideas into the workings of the bank. Such an idea was discovered in the telecommunications sector a model which was designed to prevent and predict churning of customers. The bank recruited an expert to design this model for them, but it was a total failure, as the workings of a bank are completely different than that of any telecommunications company.
Skills Failures
Poorly handling unanticipated and sudden problems
A multinational company recruited a large team of data analytics, who expressed the wish to make certain insights available to everyone in the company. This prompted for a large scale effort directed towards big data support and analysis, which could not be handled properly owing to the lack of skill of the support staff, and the absence of proper IT knowledge to get the entire analytical project up and running.
Absence of skills to analyse data
A retail companys CEO wanted to keep his banner away from the clutches of Amazon, so his CIO was asked to come up with a customized recommendation engine to assist the company. The CEO was promised that it would take about six months for the engine to be delivered, but such an engine required the usage of collaborative filtering, which was beyond the skill of the development team.
Strategy failures
Choosing incorrect use cases
The case of an insurance company trying to analyse the proportional relationship between bad habits and good habits, and how it affects the purchase of life insurance policy is quite notable here. The entire analytical process turned out to be a great failure.
Organizational inertia
A travel logistics firm wanted to discover the true nature of customer behavior by digging into the website weblog data. It was soon discovered that in reality, customer behavior was completely opposite to what the management had assumed.
