The lessons of Waste Management for Data Management

By observing the evolution of waste management, we can learn valuable lessons on data management :

• Switch from bulk collection to selective sorting

• Sort as far upstream as possible

• Organize the material courses

• Trace the material as the process progresses

• Enrich and enhance each item

• Valorize the placing of the item on the market

• Enhance associated services

The technological advances of recent years have made it possible for any company to acquire data in large quantities : The appearance of connected objects, the increase of bandwidth on the networks, shared storage, are all factors of acceleration .

The arm race has created a flourishing market around data. It is suggested to the manager that possessing the data will enable him to make the right decision.

In a short time, the company therefore has terabytes of data ... and its associated lot of inconveniences : servers, security procedures (access, backup, redundancy), etc ...

There, the problems begin. What to do with this mass of data?

Very quickly (often before the end of the initial implementation of a software ... like an ERP system), the manager may feel the need to address new issues, such as putting a process under control, monitoring an emerging market, Implementation of a new grid of analysis of its costs, or the change of ERP.

How was the data organized ? Who carries the responsibility in data management, and the whole model ? How are data classified, monitored, stored?

The result is that the company has thrown away all the things it could collect in one or more warehouses ... and finds itself with a lot of rubbish (a "nice ocean of data" for poetic marketers).

So :

• Switch from bulk collection to selective sorting : don't collect data to collect data. This quickly becomes a financial burden, a technical complexity and makes you incapable of any use of this data.

• Sort as far upstream as possible : But for that you need to know where you go. You must first define your business strategy, and than align the data management policy. It is a business task, not an IT activity. Become the architect of your data governance. Draw your model.

• Organize the material courses : Reassemble the current, from your final need to the data you need for it. Define your referential data, each responsible for each data, associated procedures, from collection to diffusion. Decide from your sources, from your data providers, the expected accuracy. Leave aside everything else.

Surround yourself with IT designers to urbanize your IT model, who will define your flows, avenues, streets, warehouses, data enrichment workshops, and will oversee the IT specialists who will build up your model. Build your models as neural networks,

• Trace the material as the process progresses : Define your indicators of quality, enrichment of your data and your models. Put them under control. You have to know where are your "raw material", your "in-progress" item, your "final item". You have to monitor the process too : number / types of failures, etc ...

• Enrich and enhance each item : Think about the future and type your data with possibilities of combinations, enrichments. Build an evoluting models capable of evolution. A data is not available today? Prepare him a place in the model and share your need. Engage providers, researchers, groups. Organize hackathons.

• Valorize the placing on the market of the item : Because at this point, the data has a real value, choose your model of diffusion ... or protection. If you want to do business with the data, put yourself into trading and put it up for sale at the best time, or generate recurring income.

• Enhance associated services : Finally, offer services, advice, around the given since you master the model. Support your customers or colleagues. The feedback they will give you will enrich you in return to improve your model, its efficiency.

Conclusion : Data management is primarily a matter for professionals. IT experts, Towercontrol solutions and Big Data suppliers only have to support you strategic schema.

And perhaps the actors of ontogology that appear actually will be the missing link between these actors.