Making Big Data Work

Making Big Data Work

Fri 14 Oct 2016

The unique business transformation attached to the digital era requires companies to respond with velocity.  New systems integrating core transactional assets with mobile and social media have to be used – implying their ability to face volume. Moreover those new systems need to manage enriched operational reality and risks.

Laurence Malroux, CEO and President of Big Data Analytics-as-a-Service firm, Panoratio, talks to Mazars about the opportunities and challenges of Big Data implementations. She specifically explains how big data analytics business applications can reduce the complexity of making big data work for the benefit of business: i.e. providing operational improvement and productivity.

4 Questions to Panoratio

What are the challenges of implementing a Big Data analytics application?

Laurence Malroux: The challenge is threefold. First, the raw data itself is a political and organizational challenge. The right granularity, depth and volume of data need to be collected from all departments and applications concerned: this means inside and outside the company. This requires a secure and easy architecture without any department fearing the impact on their activities. We know that Big Data can unlock significant value by making real information interact with each other. Most of the time, business people are overwhelmed with the current accumulation of aggregated or pre-calculated data that does not provide any meaning: the ability to get all data – including all details, is essential to get the real value from data.

Second, data connection should not be confused with data combination and communication. It is the combination and communication of data between itself that provides the most relevant value. The meaningful story is achieved through the communication of all combined data. These interactions link to operational decisions and can be tested before any implementation of a new IT project. Monitoring the impact of any action from as many angles as possible is not only wise, but economically preferable.

Third, data timestamp reality is also a key factor. Past data feeds predictions and, coupled with present data, drives prescriptive actions. Prescription is the key feature for Big Data analytics application users’ productivity and adoption; it requires a full detailed understanding of many business angles. The volume, variety and velocity attached to Big Data can be the business users’ nightmare: more options, more choice, more complexity… Happy Big Data analytics application users require plug and play customization, which is the holy grail of Big Data analytics application.

How can independent strategic analysis help with such challenges?

Laurence Malroux: Reconciliation of all accessible and often unknown data can provide business experts with the real information they need to identify the essential trends of their business, and clarify the strategic decisions that have to be made. This is a prerequisite to identifying key performance indicators and implementing an associated data monitoring system. But as described above, it can be a complex and lengthy process. However, an independent strategic analysis of data can not only speed up the process, but as it relies on a simple copy of production data, the process is not invasive and minimizes the involvement of IT departments. This all helps to facilitate decision making required to manage operations over time and ensures that business owners keep control over their project.

What does independent analysis involve?

Laurence Malroux: A sophisticated big data analysis engine handles large volumes and variety of structured data to perform on demand, complex, in-memory analysis on multiple large data sets. Analyzing historical and transactional independent data results held form new patterns and allows the identification of risks and opportunities. The speed of processing ensures a quick diagnosis so you can use the service as a prospective discovery tool to identify trends before scrolling down into in depth analysis as required. Analysis results can be provided on-demand through multiple reports as required, from simple spreadsheets to sophisticated graphics. Based on how we work, analysis does not require any installation on customer premises as services are provided through an SaaS online service.

What does Big Data Analytics Application involve?

Laurence Malroux: We know business people are overcrowded with data – so only a relevant and precise business-oriented application will make the difference.

In the end, it is not the data that businesses are looking for, but what they can learn from it and focus on thanks to its findings; letting models and algorithms (AI/Machine Learning) do the bulk work part. This is why it is so important to build applications that support main business goals and fit with specific business issues rather than providing additional analytics platforms that are just more noise by providing even more unprecise and incomplete results. This is the condition to build an application that really matters to the business, rather than providing additional data and analysis to an already existing ocean of data and analysis without measurable operational gain.