The lessons learned, so far ...

Earlier this week I turned 50 years old, and I have worked with evaluation, data and analyzes for half my life. It remains an exciting journey - especially where decision intelligence is used as strategic management information, but also the professional meeting between the classic social science statistics and the modern data science analytics.

My analytical career journey at a glance

As student at the university I started with mainframe programming Pascal and SAS back in 1989. Later at my first job as analytic assistant in 1994 I used Lotus on datasets on 5? " floppy discs. I was happy when we had to move on the new frontier to programming SPSS, Lisrel and Amos on Personal Computers and when we learned to store the data at a network drive. Later I used Html, Php, Css, Excel and QlikView to make nice data visualizations. Today we work with big data in cloud-based data lakes, and I needed to learn R and some new tools to handle the modern environment for the data workflow from raw data to presentation of the results. Now as Head of Analytics I have to care as much about the policy for DevOps as how we make the metrics useful for evaluation of the strategy.

I do not think a specific software is important. However, it is important to use more open source. Anyhow for me the tools is just symbols of a thoughtful long journey with my hands deep buried into the data in a changing world.

Why does 25 years of experience matter when new things are constantly being developed to the market today?

External consultants, internal analysts, executives at all levels have a very volatile memory in this accelerating age of time. The development for particularly quantitative analysis, such as advanced analytics, is really a journey in the fast lane, but I find that many analysts forget the classic virtues of methodology in their analysis work. Especially during the current corona crisis, astonishing many people have suddenly become a form of self-proclaimed epidemiological analysts - please take care out there!

This article is not the approach known from “The World of Yesterday” (for you remembering Stefan Zweig) – I am still curious and active learning how to work in the field, no longer as analyst doing data wrangling, but more with a strong focus on how different experimental, explorative and explanatory analytics can be used as relevant data intelligence for executives data-based decision-making processes.

The famous unicorn team:

Doing useful analytics requires complex competencies.

When I look at my team we must deal with very different kind of skills and competencies to achieve success as the central strategic intelligence team at the university. We need to work close together and share all our knowledge, not only internally in the team, but very much outside our analytic team. There is an on-going increasing complexity, and we know from experience that no external consultant can walk the way for us – there is no quick fixes. This is an expedition like Shackleton polar voyage. I am really proud of my brave team members.

The first lesson: Humanistic abilities are important prerequisites

We must stay curious to the maybe more or less blurred strategy processes, the policy development in the context and the changing external political interests to make the most relevant analyzes for the executives. And we have to handle project management, planning, implementation, follow-up and evaluation to be effective. Add to this that we persist to think about ethics, legality, user involvement, and building organizational data literacy. Do not forget this is some of the softer and very important competencies that make your team wins in the long run, actually also on the short sight these skills promote and constraints your settings for the harder data coding process. 

The second lesson: There is always a context and a target group

We really need to have a strong focus on dissemination and communication of numbers, figures, graphs, statistics and how the executives get beneficial use of the analytical perspectives. Stay in dialogue with all stakeholders in the surroundings. Reporting as management information (ie. Business Intelligence) for compliance or strategy needs a lot of love – be careful and transparent about definitions and the data models behind the results. Anyhow you have not finished your work when you get the results. Data visualization, data storytelling, making infographic and dashboard design is most important for our decision intelligence. Focus on the decision-making process when we present the results are crucial for our raison d'être.

The third lesson: Consider carefully your approach to statistical analyzes

Both the classic social science approaches and more modern data science approaches (eg machine learning) has advantages and disadvantages. Today everything moving fast ahead, I myself also experienced that I sometimes also has been running too fast. We really need to be more aware of what kind of approach fits the purpose: Explainable analytics? Explorative analytics? or maybe Experimental analytics? Well, I do care about the classic virtue in statistics to perform prerequisite tests for the models and statistical tests used. We need to know the bias, outliers and weakness in our data. And visual evidence is not real evidence, and there is a critical difference between co-variation and causation. At this point I have no problem being an old grumpy man, because this seems to be a surprising challenge for even talented data scientists, when I read analyzes at different web-pages around the internet.

The last lesson. T-skills in the new black

The journey from floppy discs to personal computers to network storage are moving on to cloud solutions, and what do we see next? This not only a story of more data and Big Data. It is also a huge change in workflow and skills required. There is no longer only data processing, programming, data wrangling and ETL processes as the traditional and huge task for analysts. Now new tasks have shown up on the analytic office desk: The team needs to handle data engineering, eg. API, automation, as well as database processing in relation til warehousing, e.g., Azure Data Lake, DevOps, Git et cetera. This is the lesson of a very high value of more technical skills.

Exploration of the map for navigate the upset sea of data

I want to go into detail with two issues that often challenges analysts and data scientists because it is not their main focus. And it not so sexy as web-scraping, API and machine learning, but I believe the analytical fields has two trending issues:

• The responsible use of data.
• The dialogue with the organization brings the numbers to the table.

The responsible use of data.

Data ethics is not a new dimension, but it is still important to handle. For those of us who have also worked extensively with qualitative methods and evaluation, we know it can be very an inherent basis to have the ethical perspectives integrated to the methodology. Too many executives and data analysts today are too busy focusing on how far you can get data to show competitive advantages, without having enough worries about the values of the object field, the people involved, derived effects et cetera. As well as to few remember to discuss what the data can't be used for in the situation.

It's not just about the ethical dimension, it's also a broader perspective about taking responsibility for using all data wisely. And of course, data must be used to give us new and necessary - and perhaps unpopular or even unpleasant - insights. On doubt. And there is always more data to add. On the other hand, I had experienced how GDPR been used almost like a religious revival to stop analytics projects, which in my view is also a huge misunderstanding. GDPR is not a tool to curb the use of data, but a method of ensuring that we work wisely, well-informed and responsible with the data.

In my opinion data ethics does not slow down the development of new fancy analytics as example predictions of individual behavior. The ethic perspective ensure that we are going on consciously and reflected on the use of (big) data.

I am proud to have helped develop our new Forum for Educational Data Ethics at University of Southern Denmark (SDU), where representatives across the University meets - this is for researchers, administrative staff, analysts, lawyers, the DPO, and of course the students. Here we have a possibility to discuss ethical perspectives on the use of student data.

In various contexts, I see this classic virtue of responsible use of data is a challenge. I think we should all give more awareness to this issue in the future.

The dialogue with the organization brings the numbers to the table.

Numbers, Statistics, Analysis has never in itself been of any value. A certain number, an arbitrary graph, a comprehensive dashboard or a colorful infographic do not do the magic trick for use without preparation and presentation. And not only do you have to keep the target group in mind, remember to include all or most of the stakeholders.

When my strategic intelligence unit produces the top management information to the executives, we simultaneously make the same analyzes available to all employees. We write about the new analyses on the intranet, and we present our new findings in all the meeting contexts we can reach at the university. Often, earlier in the process, we have been in an informal dialogue with the organization about what are the most appropriate choices analytically with respect for the local / specific domain knowledge. This can open for a broader analytical approach, better data, more relevant metrics and might promote the use.

In the end it’s all about the use of the analyzes. Therefor I have focus on continuously developing data literacy. My ambition is that all kinds of colleagues achieve an increased data savvy and analytics capacity – I even have developed an analytic course for none-analytical-colleagues to support this journey. In this way we can create a collective critical thinking to develop our overall organizational analytical readiness. Indeed, this is most important in an increasingly data-driven world where all numbers do speak, if they are tortured long enough.

Stay in touch - Learn & Share

First started you quickly realize there is no end to this journey and your coding skills in a particular software environment is not enough to bring you to success – In my view handling data is the key necessary skill, but it might be the least of your issues. So, stay open minded, be aware because around the next corner there is a lot more to learn. Keep your relationship with the organization tight with the communication lines open. I myself are ready for 25 years more in this business – In enjoy the ride - Lets stay in touch :-)