Many entrepreneurs and managers are aware that the data available within their company can be a valuable source of information. Some of them look at that data as a real goldmine. But when it comes to the successful extraction of that digital gold, there is often much more to it than people think.
First of all, it is important to find the right data: knowing what you have to search for and where to search for it, because that data is often unstructured and stored in multiple ways: in a database, as Excel files, … It can also be photos and videos. All that data can usually be found in different places as well and is usually not accessible for everyone.
Ask yourself: what do I actually want to be able to predict?
Companies rely on our services to obtain more valuable management information. In order to achieve that, we go through a number of phases together with the client during which we, as Amsterdam Data Collective, can provide various services.
It starts with the question: what information would you like to obtain? What is the phenomenon you want to predict? For example, the manager of a portfolio of financial products wants to know who the risky customers are. It is useful for a marketing manager to find out which existing customers are most likely to generate extra revenue. It’s basically about asking the right strategic questions and often comes down to: where can data help us be more profitable?
Question 2 is what data should be used to be able to gain that profit. Then you look at a combination of internal and external data: we combine the data available within the organisation itself with valuable information that is available elsewhere in the world, if possible.
Management decisions need a reliable basis
Once all relevant data has been traced, validation is an important step. By cleaning up and structuring the data, and putting it in the right format, we make sure that the input is correct and that we make that input useful as well.
In the next step, we develop an algorithm that processes the data in the right way, so that it is possible to make a reliable prediction. If that works, we turn it into a dashboard on which all relevant management information is presented.
Clear and transparent, so that the initial question with which the process began (who are the risky customers in our portfolio, which customers provide the greatest opportunity to achieve more revenue) can be answered. Do it in such a way that it becomes an interactive process and that it keeps being updated in real time. That always ensures the most accurate basis for management decisions, also in the future.
You have to be honest about what is possible AND what is not
We help our customers with all phases specified below. Our services can be divided into:
- advising about a data analysis strategy
- developing and implementing algorithms and models
- validating and improving existing models
- automating data analysis and reporting
Often, it is a combination of two or more services. We are clear about what can be expected of us. In my opinion, you should always be honest about what is possible AND what is not. We choose for a practical approach, but never without relying on a scientific basis. For example, we indicate how reliable a prediction is. So not: ‘Next year employment will increase by 1 percent’, but: ‘… between 0.8 and 1.2 percent’. And then I add: ‘with X percent certainty’.
I’m convinced that our people are able to extract all relevant information from complex data sets – AND that the information is correct. Only then can our clients make informed management decisions that fit within the vision of the company.
Whether you can call it a gold mine… I think it’s just smart use of your organisation’s resources. That often turns out to be more than what anyone of the board would have hoped for.