The world’s leading publication for data science, AI, and ML professionals.

Why the world needs datapreneurs

Insights from a personal story of why data initiatives may have failed in the last decade. From Data Product Management to Datapreneurship.

Insights from a personal story of why data initiatives may have failed in the last decade and still are challenging as of today. From Data Product Management to Datapreneurship as an approach to holistically tackle digitalization of the future.

After working nearly a decade in the field of data science with its respective domains such as business analytics, business intelligence, advanced analytics, machine learning as well as artificial intelligence, I finally had to accept that creating value from data is – and probably always will be – a challenge. While I was delivering hands-on data science solutions in the first few years of my career, in later stages, I co-founded a company called Mindfuel and began to study and analyze the organizational ecosystems and environments, into which so-called data initiatives are born, from a more strategic perspective. This is what I have found out so far.

The data science wave of the 2010s

Ten years ago, the European market was in the hands of business intelligence solutions and their providers. In most cases, data was historized in relational databases and SQL was the go-to programming language to massage them. At the end of a day, a business report was generated or a dashboard was deployed on a server to which a business user had access. Thanks to some big players, a new discipline was about to swash over the ocean from the US to Europe and its name was "data science".

It was the beginning of a hype. People wanted to use advanced methods to get more out of data – more information, more insights, better actions – simply said: more value. A new demand emerged. We analysts learned to develop models, create showcases, build prototypes and even deploy first pilots. Proud as hell, we demonstrated the company’s next big lighthouse projects to executives, and they were excited and delighted at the same time. Moreover, they provided us with further financial funding and budgets to keep going.

Their expectations were implicitly clear: Please deliver value from these initiatives. Some of these stakeholders invested heavily. They financed centers of competence, data labs, AI hubs, data units, and of course, all necessary technologies to build a state-of-the-art ‘data universe’. CDOs were hired to dig the gold. The market was growing limitlessly until one day, the first stakeholder asked a magical question: "When approximately will I get a return on my investment?" From this moment the mood tilted.

After realizing that only a limited number of data initiatives was really paying back on all the efforts or becoming break-even one day, well-known research institutions jumped in and explained that ~90% of all data initiatives won’t make it into production. We all had to admit that developing the technical solution worked somehow, but was not enough to deliver ‘the value’.

Finally, the hype flattened slowly towards the end of the decade and today’s decision-makers started to ask for an ROI concept first before they approve budgets and investments.

My observations why we failed so far

Yes, we failed. We failed jointly: experts, vendors, users and clients in terms of organizations with both their employees and stakeholders.

Here is the sneak preview: Technology was not the one to put the sole blame on. Within the last ten years, we improved in designing and developing better models, better infrastructures as well as better platforms – or to sum it up: better solutions. Thanks to the efforts of existing and novel providers and open-source communities, it has never been easier to execute a regression model on a dataset or to scale out a model, which needs more computational power with a simple click.

However, somehow the respective transformation of an appropriate organizational environment was missed. My major observations are summarized as follows:

  • People pretend to be agile but still think in waterfall mode or at least struggle to put agile methods into practice
  • Organizations still lack a proper data operating model to develop data products and a corresponding process of how to integrate these data products into a target ecosystem
  • The data world, in general, just seldomly considered the user’s needs in terms of desirability as of today. To this day – the main focus is the feasibility
  • A culture and mindset challenge: how can we collaborate among business units – especially when it comes to data
  • Focus on output, not outcome: What is the difference between output and outcome? We execute on projects all the time instead of developing products, don’t we?
  • No one is responsible for the business case and ensures that there will be a proper return on investment

All these observations have nothing to do with well-known challenges within data science such as how much data is accessible, available, or ready to be analyzed but go far beyond the data space.

Learnings from digital product management

Somehow frustrated by the status quo, in 2019, I presented my findings to one of my best friends and today’s co-founder at Mindfuel Max, who worked as a digital product manager at the time. He instantly encountered some untenable assumptions:

  • You do NOT develop products, but projects.
  • You do NOT consider desirability when developing your solutions.
  • You do NOT consider marketability since your solution will compete on a market against alternatives.
  • Who owns the viability of your solution? Yet, it cannot be the data product owner.

My mood turned from frustration to rage to motivation. He does have a point! When he explained to me the core of product management in greater detail and the role and responsibilities of a product manager in specific, it rang a bell and raised a reasonable number of questions: What does this all mean for a data science project? Could a product owner or project lead of a data science project be the person taking the responsibility for the ROI? Or lies the responsibility with the typical business owner, who provides requirements and domain knowledge to the project? However, is s/he in the right position of owning the entire solution? Isn’t s/he the recipient of the solution in the role of a user at the end? If so, who finally owns the data product? Why does a proper product manager role not exist in the data space?


Mindfuel's Approach for Data Product Management
Mindfuel’s Approach for Data Product Management

Keeping the value creation promise to stakeholders in mind, Max and I soon after entered the next step of this journey called Data Product Management. Adapting methodologies and frameworks from (digital) product management opened up a whole new perspective on how data initiatives could be realized and delivered in the future. We blended approaches and concepts of agile frameworks, execution engines and product management tools with artifacts from CRISP-DM, ML- & DataOps and topped all of this with principles from the data strategy domain. This resulted in a holistic approach greater than the sum of its parts.

Why the world needs Datapreneurs

Yet, all our efforts led to the fact that there may be more to add to this story than just Data Product Management. We realized that data product management is part of a discipline called Datapreneurship.

The Datapreneur is this kind of "egg-laying woolly milk sheep" the data scientist used to be a decade ago. A captain in each initiative holds course while the ship is sailing at full knots. A person who accepts the risk and is eager to deliver the promised return on investment. S/he is the person who owns the vision of a product greater than the ML model behind it. S/he keeps users’ desires in mind at all times and creates business-critical success accordingly whilst understanding and tackling the challenges of the data world in practice.

Definition of a Datapreneur
Definition of a Datapreneur

But from a more comprehensive perspective, it is more than this:

It is a character that has the guts to fail, quickly learns from it and creates an ever-growing life cycle out of this journey. A cycle where innovation and value creation go hand in hand. It is a person that respects the fact that investor’s money is no free lunch but needs to be paid back with a reasonable multiple considering not only business’ interests but also morals and ethics, especially when it comes to data and its potential dangers of misuse. A Datapreneur always truly cares how data is used in a product today – and will be used in the future – while considering the product’s greater cause. A Datapreneur delights the world with innovative and sustainable data products people love. Every product s/he designs and builds makes its environment better – step by step, idea by idea, product by product.

One day, the Datapreneur shall be part of every data initiative ensuring value and success from digitalization.

What are your thoughts on this? We on our part will continue on our journey to finally decrease the percentage of data initiatives failing and we look forward to any companion along the way.


Related Articles