First, it is good to clarify what is meant, when discussing artificial intelligence and other forms of advanced analytics. A data-driven business utilizes a wide range of technologies, not all of which can be bundled under the term AI. It is often more precise to talk about advanced analytics instead of artificial intelligence, as advanced analytics is the umbrella term for all the different approaches used in a data-driven business. It is
a prudent to save the term AI for solutions where the intelligence of the solution is a reality rather than mere marketing rhetoric.
• What goals should the organization set for itself and its partner?
• What prerequisites must be met prior to implementing advanced analytics (and artificial intelligence) solutions?
• What are the short- and long-term objectives – and why both are important?
• How to ensure that the investment delivers value as quickly as possible?
With this guide, we want to increase companies’ understanding of advanced analytics, as we believe that technology will only be useful if used properly.
We will discuss issues that an organization
should consider before investing into technologies or committing to development projects:
Organizations and their needs, as well as ope- rating environments, are unique, which is why detailed generalities cannot be made on the application of advanced analytics. However,
we hope that this guide will help you refine your vision of how data-driven business is done in your organization.
Enjoy the guide,
Nicola Holmes, Senior Data Scientist
Eerik Puskala, Partner
No development project can be started without first looking at the current state of the organization. Data-driven business should be approached holistically, and thus the prerequisites for a data-driven business need to be assessed holistically as well.
A development project needs to have clear objectives. Objectives give the project direction, and they guide the decision- making during the project. Excessively detailed objectives coming from top
down, however, should be avoided.
Advanced analytics solutions can be separated into different levels. The higher the level of the solution, the greater the business benefit. At the same time, however, the requirements for the user organization scale up as well.
But regardless of the level, a properly executed advanced analytics solution will always be benefi- cial for the business.
If the solution can effectively generate reports and views that show you what has happened and why it happened, the solution can be categorized as descriptive or diagnostic data analytics.
If the solution, on the other hand, can generate reports that tell you what is going to happen and in what statistical probability, or it can simulate
different scenarios, it is called predictive data analytics. The business benefits of predictive analytics are greater, but still limited if the analysis does not give more details on the cause of the predicted outcome.
Imagine a large piece of manufacturing machinery that can tell you that it is going to fall apart, but not the cause of the failure.
The highest level of advanced analytics is called prescriptive data analytics. It refers to data analytics that tell you what is going to happen and why. That is, what choices need to be made to ensure or avoid the implication of the forecast.
Imagine a large piece of manufacturing machinery that can tell you that if the cooling of its control circuit board is not repaired, the device will stop production within 12 hours.
The degree of maturity of an organization deter- mines what kind of advanced analytics solutions can be developed in the short term and what will only be available in the long term.
In this context, maturity refers to the organiza- tion’s overall readiness to utilise advanced analytics. For example, the implementation of descriptive analytics solu-tions may require:
Technology such as data warehouse and a BI solution.
People with the ability to build a descriptive analytics solution.
People with the ability and willingness to use data to support their decisionmaking.
These requirements are not unreasonable in the short term and the business benefits of the solution are tangible.
The development of a prescriptive analytics solution, however, requires much more.
A vast amount of high-quality data in a continuous stream.
The solution will likely require its own technological environment.
Versatile expertise in data, data analytics, architecture, machine learning, application development and business areas is needed.
Data management must be integrated into the organization’s operating model at all levels. Change management is therefore also necessary.
Such solutions are typically unique, and their development is slow. Various experiments and failures are a part of the development process. It is quite possible that the level of maturity regarding technology, data, processes, and competence must be increased before it is possible to start an actual development project. And when a development project is started, it is worth starting small.
Before making investments, it is good to be aware of what kind of advanced analytics solutions can be deployed quickly and which require more time and resources.
The maturity issue is an inevitable part of the transformation towards a data-driven business, and it should be addressed early on. A good partner can help both define and develop your organization’s level of maturity.
The chances of success increase significantly, if your development projects are backed by a clear and cohesive vision shared by the entire organization. The top-level vision guides the way things are done and ensures that the projects have the support of the management.
The vision is not enough for successful development projects to be carried out, but it reflects the commitment of the organization’s management to develop the use of data and support the initiatives. In our experience, many organizations have, by now, a vision of becoming a data-driven business.
A top-level vision should lead to goals for other levels of your organization, such as sales and production. In this case, the challenge may be how to get other levels of the organization to adopt data orientation as their goal. Therefore, change management at all levels of the organization is an important part of the transition into a data-driven business.
When preparing an actual development project, the goal may be, for example, to improve sales efficiency through better product grouping and customer segmentation and by improving sales predictability. Or the goal may be to streamline the production process by improving quality control through data.
Agile application development uses the term “EPIC” to describe major development lines that are broken into several smaller sprints. This is also applicable in the development of advanced analytics. EPICs (or use cases) are ideas how data could be leveraged to benefit the business, and a capable partner helps to convert those ideas into smaller actionable steps i.e. sprints.
Data-driven businesses and strategic visions are fun to discuss. However, in order make sure the talk gets turned into action, the management of your company must genuinely commit to data-driven transformation.
This means that:
management must actively promote change in their own organization.
the project must have a business owner, who is actively involved in the cooperation with the partner.
adequate resources must be made available for the development project.
Technological development projects often fail due to a lack of a business vision. This is because, although, technology specialists, data scientists and analysts are experts in their field, they are not experts in making the best choices from the perspective of the business.
Within the context of agile application development, we use the term “product owner”, who is someone that acts as a representative of the business and the users in the application development process. The same practice is applicable in advanced analytics development projects.
The role of the organization’s own level of competence must also be emphasised. For a company to be able to effectively convey a business perspective to technological experts, product owners must accumulate sufficient expertise in technologies and opportunities provided by data. This allows for effective cooperation and helps to avoid misunderstandings.
In the initial situation, the organization’s own level of competence may be relatively low, but its development and expansion should be a priority. Expertise is best developed by doing and experimenting.