Not so scary audit...
Audit. How we dislike that word. It has such a negative connotation. It's as if we already assume in advance that something has been done wrong. This is probably our experience, because the audits we know usually end with a list of remarks, shortcomings or mistakes, and don't really contribute anything more. But - from my perspective - a good audit is the sum of three elements: expectations, reality and a plan for the future.
I am writing about this issue because I have noticed that many organizations would benefit from an audit of their data processing and usage ecosystem. Why? We say that data is extremely valuable to us. That they are like gold or like the new oil. But on the other hand, we most often deal with this valuable resource chaotically and without a plan. And the way out of just such a situation is to conduct a data ecosystem audit consisting of three key steps.
The first step is to set expectations for the data ecosystem. This is a task essentially abstracted from technology. It's about identifying both reporting needs, the availability of data for statistical and visual analysis, the ability to make data from a variety of sources available directly in systems such as a CRM or Customer Service system, and the ability to build a variety of machine learning models.
In other words - this part of the audit involves brainstorming across the organization about current and future opportunities to use data to improve processes, increase efficiency, enhance services or even create new ones. After this stage, then, we have complete information on what the organization would like to do with the data it has or can acquire.
The second step is to confront the reality of the data ecosystem. We start by identifying the organization's existing data sources, both those related to specific IT systems and all other significant sources, such as Excel files.
Next is to look at the system for processing data into analytical form. This raises questions about the existence of a data integration system in the organization, such as a data warehouse or data lake. The key here, of course, is both the scope of available data in such a tool, i.e. the question of its completeness, and the very execution of this system.
At the end of this stage, it is necessary to verify how today's data is used in business processes. Are there reports in the organization and what is their performance? Are there machine learning models and is their accuracy and performance satisfactory? Finally - do marketing, sales or customer service support systems provide employees with the complete and high-quality data they need in their daily work?
Plan for the future
The third step in the audit is the intersection of business expectations and existing reality. Such an analysis shows differences on two levels: business and technical. The business aspect is those expectations of our employees that would help the development of our Insurance Company, but which have not been realized so far. Maybe we are missing certain reports? Maybe some processes could be faster and more efficient with machine learning models? Or maybe we are simply limited by insufficient processing capacity or poor data quality?
The technical aspects, on the other hand, look at how the data ecosystem is built. Is it standardized? Are we creating the right documentation for it? Have we used the right tools and architecture? What else needs to be added to our data ecosystem? So in this aspect, we find gaps at the level of not having a data quality management system or not having a system to run machine learning models, but also flaws in the architecture and performance of current solutions.
On the road to strategy
At the end of an audit carried out in this way, a document is produced that describes what path our company has gone through in terms of data processing and use. It explains where we are, that is, what solutions we already have in place. It further describes what we would like to achieve in the future. What we need for our business to thrive.
An audit conducted in this way becomes the ideal basis for defining and writing down the organization's data use strategy. And this, in turn, allows us to define a plan for further action, both in terms of timing and the investment budget we need. So this is a great first step on the road to building a data-driven organization.