At the end of our article entitled Adapt or Die: Big Data at the heart of business transformation, we asked two questions, the first of which we will answer today: How do you initiate a Big Data approach and what are the concrete benefits?

According to Forester Research, there are three main reasons for the failure of Big Data projects:

1. They do not start off with a question
2. The technical skills and expertise required is underestimated
3. Another data silo is created

Points 2 and 3 are not all that surprising, but what about the point 1? After all, isn’t the goal of Big Data and analytics technologies to find a needle in the haystack?

When using a search engine, for example, the first question gives results that allow you to refine the search, and then you repeat the process until you get the desired information.

It is the same for Big Data. Without context, you cannot know if you're going to simply get noise or a clear signal.

Start with the right question

The first step is to define a clear question that Big Data and Analytics technologies will be able to respond to. Suffice it to say that this cannot be an open ended question such as: How can I increase my sales? Or will this particular product succeed?

It involves precisely defining the query in terms of the nature of the current challenge faced (eg risk analysis, cost control, process efficiency ...) that cannot be resolved with the infrastructure and tools in place. Then consider whether adding additional data sources will help to achieve better results.

Use the right criteria

To determine the type of question you want to ask, it is recommended to use the following criteria:

1. Will the answer to the question require additional data sources then those in place?
2. Will the answer to the question give usable results?
3. Will the answer to the question provide a sufficient return on investment to justify the project cost?

Define the project scope

In addition to asking the right question, you must define the scope of the project. According to a survey by Infochimps, an inaccurate scope is considered by 58% of respondents as a main cause of failure of their project.

The role of the multidisciplinary team you have put together is to define the project's objectives and its scope and plan the Big Data project, including the sources of data you will need.

Build a use case

The best approach is to create a use case that defines the company’s question and serves to support the proof of concept. Here are two of the most common use cases, which can serve as a guideline.

What can I learn more about my clients?

It is an excellent starting point, allowing you to use the data from sales, accounting and other departments, as well as external data.

Companies can also monitor consumer trends using reward points, surveys, social media, web requests and other data sets.

How can I improve operations?

You can use the data generated by machines and operations, in conjunction with those of sales, to create real-time transactional analysis of business processes.

Companies can thus make substantial savings by integrating Bid Data solutions to manufacturing, and by using data from factory machinery, in order to reduce manufacturing costs and improve quality control.

Conclusion

Once you have defined the challenge you want to rise above and have built your use case, it is now time to assess the available resources to determine what you already have and what you need.
This topic will be addressed in our next Big Data article….stay tuned!