Big Data + Wrong Method = Big Fail

Big data is hyped as a gold mine, but Big Data applications are risky. Understand how to start with a minimum viable application and iterate to minimize the risk of failure.



By Abed Ajraou, Head of Data at First Utility.

It’s official, there is no longer any consulting company that has not started the conquest of the Eldorado of Big Data. Indeed, recent analyzes have shown that the Big Data market would generate $ 40 billion for 2015 and it will grow by 14% every year until 2020. Race to the huge mass of opportunities became obvious to these companies, even those who declared a short time ago that Big Data was only a buzz.

However, rare are the companies that support businesses in the best possible way, ie by offering an innovative method and adapted to the challenges of Big Data.

Navigate in the unknown

Big Data + wrong method = Big Fail!

A common feature of Big Data initiatives is the unknown. Indeed, for any company, it is impossible to predict whether the issue raised business will bring a return on investment or not.

This situation is well known by startups. When an idea emerges, it is difficult to predict if the idea will be successful without realizing quite expensive marketing researches. That is why many startups have launched a method to minimize costs of implementation and try to concretely identify if customers are interested. « Learn Startup » is more than an agile method, it sets the MVP, “minimum viable product” and iterate while analyzing the feedback and determine if the product is on track. The most emblematic example is Zappos.com; this company wanted to launch an online shoes sales platform. Instead of starting a costly development of an e-commerce web site, they defined the MVP as a simple web page with photos of shoes and they followed the return of marketing campaigns to see if customers were interested. Zappos.com generates now more than a billion dollars in revenue.

Currently, many service consulting companies advise to apply old methods (waterfall, hybrid, SCRUM …) to these new challenge. How can we ask business entities to draft a statement of requirements, or even a “user story” whereas we don’t know yet if the idea is relevant? How can we minimize risk and help companies refine their idea to become a success?

MVA: Minimum Viable Application

Unfortunately, many companies have taken the Big Data in a wrong way. They only considered Big Data as a technical issue, while it makes only sense if there are business interests. So Hadoop clusters were used to store data, even if the return on investment on the exploitation of these data has not been demonstrated yet.

In the digital transformation, it becomes vital to ask the right business questions and try to solve business problems through the data.

Like the « Lean Startup », when one wishes to launch a “use case” on Big Data, it is vital to define the “MVA” Minimum Viable Application. The MVA has several benefits:

  1. It minimizes investment while reassuring about the expected business goal.
  2. It gives business users a working and usable implementation, although it is not optimal and is still minimalist.
  3. It gives measurable elements that will lead to a new and richer version, more adapted to the user experience.

This method is intended to be iterative and cannot fall into a long and exorbitant project. Many consulting companies wish to apply the only methods they were used to. But big data is a new challenge on multiple levels: technical, business, organizational and also methodological. So beware of companies trying to sell you projects with unsuitable methods that will lead to considerable tasks and probably ends with frustration.

Original article.

Bio: Abed Ajraou, @AAjraou, is a Head of Data at First Utility, where he works on bridging business values with data analytics. He is based in France.

Related:

Original.