Abstract background

8 effective ways to make data science work for you

The experiences, ambitions and feedback of dozens of professionals who immersed themselves in data science give a fairly consistent picture of the issues on the shop floor. Eight success factors emerge. The main takeaway: let the data work for you instead of the other way round.

Key takeaways

  • Do not be blinded by the hype around data science and AI, and do not get bogged down in definitional issues.
  • A baseline measurement shows where the organization stands on the data science ladder and whether there is an improvement mentality.
  • Poor data quality is the main reason for holding back, and sometimes also an excuse to avoid controversial algorithms.
  • Data science needs a sound business case: first find out which problem it actually solves.
  • The FAIR principles (findable, accessible, interoperable, reusable) are the basis for responsible sharing of data.

1. Do not be blinded by the hype

In the Netherlands, government, business, umbrella organizations and universities try to outdo each other when it comes to investing in data science and the key technology AI. The peak of inflated expectations has been reached. In 2019 the Dutch government presented a Strategic Action Plan for AI and spoke of investments of 2 billion euros over seven years, to be raised by government and business together. Because of the vagueness of the plans, the announcement resembled the fairy tale of the emperor’s new clothes. Meanwhile business managers wrestle with theoretical and practical questions, such as how data science relates to machine learning and artificial intelligence. Do not get bogged down in those definitions.

2. Determine the ambition and maturity level of your organization

A baseline measurement is a proven means of determining how the various disciplines view data science. It is usually the first step in the improvement process: where do we stand on the data science ladder? Take the temperature of your organization. Are people open to new insights and technologies? Is there a shared vision of the future? Does the top think in scenarios: what risks do we run if we ignore data science? Is there an improvement mentality? Is data seen as the most important asset, and is there awareness that data-driven working is becoming the new normal? Or do the issues of the day rule, and is solving problems considered more important than preventing them? How about data literacy? Does IT live up to its advisory role?

3. Make data quality an absolute top priority

According to the majority of the professionals, poor data quality is an important reason why organizations are still extremely reluctant to start data science projects, or refrain from them altogether. Rationally this cold feet is understandable: garbage in, garbage out is a well-known big data pitfall. Organizations wrestle with structured, unstructured and semi-structured data, master data, metadata, sensor data, weblog data, unsecured spreadsheets and subjective data such as NPS scores. If you dump all of that in a data lake, guarding data quality becomes mission impossible.

The key question is: what can we as an organization do with the enormous amount of big data at our disposal, and how do we turn raw data into valuable information?

The impression exists, however, that organizations also use poor data quality as an excuse not to burn their fingers on controversial matters such as predictive and optimization algorithms, which people mainly see as black boxes. That does not alter the fact that good data quality (current, unique, consistent, complete, sound and logical) is of fundamental importance for excellent service. But more is needed for successful data science: internal support.

4. Introduce data science within the broader framework of data-driven working

Creating support for new technology and promoting user acceptance have always been hot issues. Business intelligence sometimes still has to prove itself daily in traditionally organized organizations. It is therefore wise not to introduce data science as an isolated project. Frame it within a broad, appealing and inspiring mission. Google’s mission is to organize the world’s information and make it universally accessible; Facebook originally wanted to give people the power to share. Their algorithms and trading practices are now highly controversial, despite crystal-clear mission statements. For organizations closer to home, the concept of data-driven working or data-driven steering is an excellent peg on which to hang their own mission, with better steering information as a tangible goal.

5. Always strive for a sound business case

Business managers, commercial directors and marketing managers want to know how data science can deliver a (financial) competitive advantage, while IT managers, enterprise architects, security professionals, PR managers and lawyers have quite different worries. And the data scientists themselves mainly want to experiment undisturbed, discover interesting patterns and write algorithms, unhindered by financial KPIs such as payback periods and return on investment, legal frameworks and codes of conduct. The picture is exaggerated, but data science too cannot escape a sound business case, a future-proof business model and an interesting revenue model. So first try to find out for which problem data science actually offers a solution. Look for the underlying basic need, formulate it as sharply as possible and take stock of the time, resources and capacity required.

6. Bundle knowledge and experience in a data team, data lab or data hub

As soon as data science becomes a serious business, it is time to look for synergies: scaling up, recruiting talent and sharing knowledge. Many professionals quickly run into a tight labour market: data scientists are scarce and therefore expensive. Nevertheless organizations look for ways to give data science a more or less formal status in or at the edge of the organization.

  • Data team. The composition is delicate. Avoid a monoculture by bringing diversity into the team.
  • Data hub. A modern, data-centric architecture for storage that supports analytics and AI by letting organizations consolidate and share data. Unlike data lakes, which were primarily developed to store data, a data hub is developed to share data.
  • Data lab. The municipality of Amsterdam chose a practical approach: its data lab is a workshop, a knowledge centre and an open stage for data professionals and an interested public. A place for smart, innovative and careful use of data.

Data team, data lab and data hub compared

Three ways to embed data science in the organization.

What it is

Main purpose

Data team

A diverse group of specialists

Carrying out projects

Data lab

A workshop and knowledge centre

Experimenting and learning

Data hub

A data-centric storage architecture

Consolidating and sharing data

7. Study the ethical codes of conduct and apply the FAIR principles

Sharing data is a trend, also within government. The Dutch Ministry of Economic Affairs explored how to stimulate the use of the internationally accepted FAIR principles when sharing private data for AI applications. FAIR stands for findable, accessible, interoperable and reusable. The Ministry of the Interior stimulates making government information available for reuse, among other things via data.overheid.nl and the API portal developer.overheid.nl.

Another initiative is the Personal Health Train, a metaphor for the system of agreements, architecture and implementation for the responsible use of health data in AI applications. Citizens, patients, care professionals or researchers drive the “trains” (algorithms) to put questions to the “stations” (data sets). The main concept is that data is not brought to the algorithm, but the algorithm to the data. The data stays at the source, while algorithms can still learn from it. Worldwide, dozens of codes of conduct have now been published. Although drawn up with the best intentions, the overlap is large and there is sometimes window dressing.

8. Use data science as a driver for your own career

Why would you burn your fingers on such a complex and sensitive subject? The personal objectives of professionals give the answer. The hunger for knowledge is great, but everyone also wants hands-on experience with the available tools. Many wrestle with their own role: how do you build a data team and find the right people with the right competencies? At the same time they want to grow into a connecting link, the linking pin between business and IT, and be taken seriously by management as a discussion partner at various levels. Some have a mission, such as convincing the organization of the usefulness and necessity of a data science project or creating the right mindset for data-driven working. And a few are honest about their own goal: keeping their CV future-proof.

How Diamond Decisions makes data science work

Diamond Decisions starts with the problem, not the algorithm. In business modeling we formulate the underlying need and the business case, in data analytics we take care of data quality and models, and we make sure decision makers understand and use the outcome.

Conclusion

By immersing yourself in data science you quickly gain an edge: you become the expert in the organization. But success in the organization requires more than knowledge: a realistic view of the hype, a baseline measurement, data quality, a broad framework, a business case, a place in the organization and attention to ethics. Then the data works for you.

Talk to the author

Herman van DellenHerman van Dellen
Diamond Decisions

Discuss your question with Herman

Tell us which decision needs to improve. In a no-obligation conversation we show you where your data delivers the most.

Schedule an introduction →

Step 3 of our approach

Diamond Data Analytics

A data warehouse that starts with your most important decisions and is ready for conversational AI: a working prototype within 24 hours.

Explore Data Analytics →

Frequently asked questions

01

How do you make data science succeed in an organization?

Look beyond the hype, measure the maturity of the organization, give data quality top priority, embed data science in data-driven working, build a sound business case, give it a place in a data team, lab or hub, and apply ethical principles such as FAIR.

02

What is the difference between a data lake and a data hub?

A data lake is primarily developed to store large amounts of raw data. A data hub is developed to consolidate and share data, so that analytics and AI applications across the organization can use it.

03

What do the FAIR principles stand for?

FAIR stands for findable, accessible, interoperable and reusable. The principles describe how data should be organized so that it can be shared and reused responsibly, by people and by algorithms.

04

Is data science feasible for an SME?

Yes, provided you start with a concrete problem and a business case rather than with technology. One well-defined question, reliable data from your own systems and a small diverse team are enough for a first project.

Further reading

About the author

Herman van Dellen

Herman van Dellen

Herman is a very experienced data scientist who has applied machine learning and statistics for many years in areas such as CRM, BI and data warehousing. He built his career at IBM, where he founded the Business Intelligence Services Unit.

About Diamond Decisions

Diamond Decisions helps organizations become better by enabling better decisions. We don’t start with data, dashboards or AI, but with the decision that needs to improve, and work back from there to KPIs, data and technology. We do that in four steps: inspiration and alignment, business modeling, analytics and data platform and data-driven decision makers. That way BI and AI don’t remain an IT project, but deliver value every day.

Schrijf je in voor de Masterclass