Web giants, startups specializing in Big Data, the market is full of Big Data solutions. To innovate more quickly and to offer better targeted and better-performing products and services, SMEs must, without delay, take up the subject.
The mastery of Big Data becomes one of the major elements of the competitiveness of companies. If a company skips Big Data, it will quickly loss its own market by a startup or a competing company that will exploit the available data sources and innovate faster than it does.
Indeed, with the exploitation of a big variety of data (text, videos, sounds, images, etc.) coming from a multitude of sources (databases internal to the company, external, social networks, studies, observatories, indicators, open data, etc.), companies can better identify the needs of the market and to offer new services and business models more quickly.
But the volume of data is not essential for the implementation of a Big Data strategy. The absence of a large amount of data should not, however, exempt SMEs from a Big Data strategy, because, whatever the volume they have, it has become crucial to take advantage of it at the same time. maximum in order to innovate, take market share and improve its processes. All companies are concerned with the valuation of data.
All sectors of activity are concerned
If startups were the first to seize Big Data, all the major accounts of the old economy have understood the interest of drawing from the data sources of innovation, productivity, profitability and disruption.
This is how insurance companies have reviewed their way of insuring their customers. Through the analysis of data from various sources – CRM tools, connected objects, market research, economic indicators, social networks, geographic data, etc. – they have become able to offer customized contracts to closely match the daily risks of customers.
Another area revolutionized by Big Data; the brand-consumer relationship. By collecting data from consumer browsing on e-commerce sites, purchases made online, comments left on forums or social networks, e-merchants gain a 360 ° view of consumers. Thanks to the analysis and cross-checking of all this data by Big Data tools, they offer real-time personalized recommendations to customers: promotional offers, other products likely to interest them, alerts on their favorite products, personalized loans, etc.
Other sectors such as e-health, transport or agriculture also benefit from data recovery. By crossing meteorological data, indicators on soil and air quality (temperature, humidity, etc.), satellite or drone-based exploitation photos, farmers are gaining in productivity. The industry is not left with the data from sensors, manufacturing processes or logistics.
But if Big Data is now present in all sectors of activity, it is clear that many SMEs still have reservations about this technology they consider too complex to implement. However, many solutions exist, be it tools “on premise” or offers available in the cloud. It’s all about organization and pragmatism.
Step One: Define the Purpose of a Big Data Project
In the preamble of a Big Data strategy, the company must define its objective. Thus, the first question that must be asked is the purpose of this strategy. What difficulty does it wish to overcome? It seeks to improve the productivity of a particular activity, to acquire a better knowledge of its customers, to create new services, to develop a new economic model or to improve its image.
Either the improvement relates to the business of the company and then concerns all the commercial activities, marketing, communication of the company, or it relates to the processes and focuses then on the production line, the logistics, the purchases etc., if ignoring this first step, inevitably leads to failure .
Step Two: Make an inventory of data
The second step is to make an inventory of data present in the company: data from management systems (CRM, purchasing, production, etc.), transactional data, official registers, websites, social media, data from sensors and connected objects, digital traces, studies, etc. The SME can then choose to rely only on its own data or enrich them with external information from many sources: open data, connected objects, indicators, measures, statistics, databases, social networks, etc.
Internal deployment or Cloud: what strategy to adopt?
Depending on the volume of data, the maturity of the company in terms of Big Data and its skills, several choices are available to it. In case of low volume, it can acquire a solution which will allow it to manage structured and semi-structured data under SQL. This solution implemented on the company’s server is free up to a terabyte of data and requires no specific skills, SQL being mastered by all SMEs. An SME can also develop applications that use analytical functions in the cloud without having to use all the tools.
A company may also choose to subscribe to an offer from a public or private Cloud operator or from a Big Data startup. By making this choice, the company accesses via the Internet Big Data tools (infrastructures, software and storage). There is no need for her to invest in the purchase of infrastructure and software.
On the other hand, it is the responsibility of the data exploitation and interfacing with the information system of the company. Skills are required. That’s why, for a fully-fledged SME, specialized startups like C-Radar (providing market data) or Dataiku (data analysis studio) are good partners to put the foot in the stirrup. These service providers support the company either in the enrichment of its data by external sources, or in their exploitation and recovery. Beyond the data and tools made available, it is above all the contribution of data analytics skills, via data scientists, which brings value to the company.
These data specialists are indispensable because Big Data, totally out of step with conventional data analysis based on the causal effect, is based on the correlation between data that has nothing in common. In addition, its operation is based on the execution of distributed processing on parallel servers, that is to say several independent computers federated as a single system. Depending on the volume of data, this processing may require significant computational capabilities. However, these means are available in large data centers such as Amazon, Google or suppliers like Microsoft, IBM, or specialized startups. But it is possible for an SME to derive value from the good exploitation of its data without resorting to significant computing power. That’s why I prefer Smart Data for SMEs to Big Data.
Big Data: a strategy that is built step by step
All professionals are unanimous: a Big Data strategy is built in stages. The company must proceed by iteration. It must begin with a use case, test it, and if successful industrialize the process. The industrialization of the process is generally accompanied by a growth in the volume of data and it is then relevant to implement a solution in-house Big Data.
If at first it is relevant to sign a contract with a service provider to test one or more projects, once the company is convinced of the benefits, it is more interesting to implement an internal solution to be more responsive and conduct very specific analyzes to its activity. It is better to train or recruit a data analyst whose role is to identify the data that provide information useful to the performance of the company, rather than a data scientist whose mission is to develop algorithms and statistical models that optimize business performance.
In addition, it is important for companies to organize themselves by bringing the data specialists and the trades closer together, because a piece of data is only valuable when we know how to interpret it. Information on a production machine will be readable by the specialists of the production line. In the context of predictive maintenance, they will be able to evaluate the relevance of changing it or opt for a new, more efficient machine, knowing, for example, that regulations or market needs have evolved. By collecting all these data and having them analyzed in relation to the business lines, the company makes the best decisions.
Establishment of a governance of the data
Finally, Big Data requires a good knowledge of the data collected and produced by the company. It also requires reconsidering the processes of collection, processing, storage and rethinking their organization in a transverse manner and not in silos. It is counterproductive, for example, to see two disjointed and incompatible client databases in the marketing and sales departments, while some of the information collected is similar.
A Big Data strategy requires managing and centralizing all the data produced or collected by the company. Stored in a data lake, these data are deposited without knowing their purpose. Big Data is not a static process, but a continuous value creation process. A very different approach from that used in the decision-making framework, where the data was orchestrated in bases according to the objective of their exploitation. Although the Big Data philosophy is to collect data without any specific purpose, the collection and storage processes must be finely orchestrated. The implementation of a Big Data strategy therefore requires the company to deploy data governance and consider that data is at the heart of its business.