The vast majority of companies are now on board with Big Data. They have equipped themselves with the necessary technology to be able to collect massive flows of data in real time. However, this is no longer what is really at stake. From now on, the real contest is the ability of companies to leverage and add value to the data they collect. Data quality has clearly become a key issue. As regards implementation, this is still far too random and unsatisfactory…
Just because we can doesn’t mean we should!
Before setting out to collect data left, right and centre, we need to identify exactly what we need.
Let’s say I go to the supermarket without my shopping list. I run the risk of being tempted by a whole load of products. When I return home, I may have filled the cupboards, but I haven’t met my need. A company that uses Big Data technology is like a shopper without their list. Before beginning the journey, it is crucial, therefore, to decide on a Big Data strategy and to identify your needs.
In this way, on the one hand, we are able to qualify the data and, on the other hand, we can avoid the pitfall of collecting spurious data that does little but fill up our ever so valuable storage space.
Once your needs have been identified, you must be able to put in place the tools and methods to harvest quality data. Did you know that a quarter of the data collected by companies is inaccurate  ? If we take as an example a company that is looking for a measure of its popularity ratings on social networks. Let’s say I’m Mars (the famous chocolate bar manufacturer). When I want to look into posts that refer to my brand, I must be careful to ignore the avalanche of posts about the American singer Bruno Mars. Never forget that data leveraging nowadays forms the bedrock of information upon which companies increasingly base their operational and strategic decisions. Decision makers must be able to rely on it. As such, it is better to focus on the quality of data rather than the quantity.
Another factor of data quality is the way it is processed. To avoid duplication, it is important to standardise the data. Products, services or locations can have several possible names. The city of Paris for instance has many aliases: “the City of Light" or even "Paname" to quote just two examples. Nevertheless, if processed separately, the results from the analysis could be distorted.
 The state of data quality, Experian Information Solutions, Inc.
Over 50% of employees consider that they have limited access to the data they need to make the right decisions. Being able to gather the right data isn’t the main thing; you have to be able to share it across your business. Therefore, internally, it is important to not just put data together, but to do so in a uniform and automatic manner so as to obtain accurate results. Data must be centralised and homogeneous in order to avoid recreating contradictory information and obsolete or duplicated data.
Continually updating a shared internal inventory of all collected data can provide this guarantee. It serves as a reference to minimise interpretations.
Lastly, data is not sacred - information that is valid today may not necessarily remain so in the future.
Therefore, we have to know how to manage the life cycle of data. Few companies are adopting a strategy to review, archive and delete data. Such management is, however, essential in order to keep a database that is sound and sustainable. Regular reviews of data allow you to guarantee its relevance, as well as to requalify or update it. This will avoid storage of information that is obsolete or incorrect. It is for the company to choose between archiving or deletion. Deletion must be well designed in order to avoid any loss of sensitive data, e.g. in compliance with regulations in force concerning the data, or to keep it secure.
These "basics" allow the company to regain control of the information. A data management strategy allows the company to gain greater visibility over its data, thereby encouraging the effective utilisation and protection of the data and also supporting decision-making. Control over data is a prerequisite in order for Big Data to fulfil its promises and deliver the customer insights required for understanding and anticipating their needs and for improving the customer experience.