In my previous blog, I used an analogy between the industrialisation of new products in the industrial revolution, and the process of driving benefits from data science initiatives in enterprises today. Data science typically produces a product that is analogous to a model or prototype in manufacturing. Whilst it has some value, typically, the real value comes with designing an efficient process for repeatedly producing it and delivering it to the ‘consumer’. The consumer may of course be a business process or an application rather than an individual.
There are several challenges associated with the industrialisation process:
- Data supply chain – the analytics are only as good as the data they are based on. The industrialising ‘data entrepreneur’ must identify where a reliable source of good quality data can be found.
- The source data will inevitably have quality and consistency issues which need to be properly understood.
- A process will be required that can deliver the data into the new process in a regular and consistent manner. This presents both technical and organisational challenges.
- Consideration needs to be given to how to consolidate data from multiple sources in to a record that is suitable for processing by the Analytical Model, as produced by the Data Scientist.
- All the challenges above need to be considered in the development of a coherent architecture and design, that can be repeated on a frequent, possibly real-time, basis.
The data entrepreneur must employ the skills to produce a smart solution to these challenges, which balances the conflicting pressures of speed of delivery, with the reliability of the solution delivered.
If the entrepreneur is successful in all of this, that only gets them as far as a reliable, repeatable process to produce his or her data product. They then need to ensure that the product gets to market. The vital last step is in ensuring that the newly developed product gets embedded in the organisation in a way that maximises the benefit from it.