How to become a data-driven company

Talking about a data-driven company in Italy is a bit like discussing science fiction: many administrators have hazy ideas on the subject, and the mistaken belief prevails that restructuring the company model towards one more responsive to the evolution of data flows (big data) is reserved for large enterprises only. Fortunately that is not so.

The architecture of a data-driven company in fact offers advantages in the immediate term and in the long term, in terms of minimising losses and increasing the company’s competitive advantage. Moreover its model is based on the creation of a smart data management platform, which will be discussed in depth further on, and which is applicable to companies of any size.

The advantages of being a data-driven company

The first advantage is certainly bound up with better decision-making. Having at hand tools able to read all the company’s data in real time and to cross different sources to find the information we need when we need it offers enormous advantages in terms of the exactness and certainty of the decisions taken. A concrete example: a chain of clothing shops might decide to cross very precise data coming from sales data, feedback from marketing campaigns on social media, socio-economic information about the neighbourhoods where its physical shops are located, analysis of the trend of the supply chain, and even weather data!

The second advantage is better “measurability”: that is, the possibility of measuring exactly and within a short time the result of a given company policy. Thanks to machine learning and the robotisation of plant it is now possible to derive and manage much more precise data, coming not only from the sales area but also from the production area. In this way greater integration of the company structure and a more effective organisation of the various departments can be obtained, with the possibility of making corrections in time before it is too late.

A third advantage, finally, is the possibility of sharing the same data at all levels of the company: one is no longer tied to company reports containing only the information presumed to be needed by a given department, but every person with a decision-making role within the company’s organisation chart has access to all the data, and in the form they want it, so as to avoid counter-productive choices. We must be sure that all our collaborators are “fighting our own war”, and to do that they all need to have the overall picture of what is happening in front of them.

What the architecture of a data-driven company looks like

A data-driven company is structured mainly on three levels:

1. The base, that is the store into which all the data collected converges. This can be managed either as a warehouse (data already grouped into “closed boxes” that communicate with one another with difficulty) or in the form of the so-called data lake (data accumulated in its original form, so that the individual pieces of information are easier to connect to one another and the ones we need can be fished out immediately).
2. The laboratory, that is the brain of this architecture. Here the data is analysed, linked and reordered in the form of a graph database (the data is related to other data through the meaningful relationships that bind them to one another, constituting a sort of “nodes” of a large network) and enriched using a Resource Description Framework, RDF (a model in which the data collected is enriched with indications about its properties and its characteristics and about how it can be linked to other types of data). Once this work of data management and analysis was reserved for the IT departments of large companies; now instead there are intelligent tools that carry out all the operations autonomously thanks to complex algorithms. That is the case, for example, of SDaaS, the acronym of Smart Data as a Service: a platform that lets companies upload their “raw” data and obtain in response an intelligent database (a “knowledge base”, as data scientists call it) that can be explored as one pleases and extended with new data at any moment, whether through a browser or through APIs. This delivers a great saving in time and in personnel for the company.
3. The command deck, that is the final station from which it is possible to have access to the data collected and classified, and to what has been derived from it. Platforms vary widely: from the more traditional analytics ones (dashboards, storyboards, reports …) to smart apps (mobile business intelligence), all the way to solutions that do not even require human decision-making, as in the case of analyses sent directly to the company’s machinery, which respond and adapt thanks to machine learning.

For all this, it follows that a data-driven company must equip itself with a clear and precise strategy for managing its data, which inevitably requires the implementation of a smart data management platform whose typical architecture is summarised by the following figure (to be read from top to bottom, in a path that goes from raw data to the knowledge of it on the part of the company’s managerial area).

Figure: smart data management platform architecture.

To learn more about the technological tools available for building a smart-data platform, see this link:

Smart Data as a Service