6 key data insights from our Data Connect Session

On the premises of former machine factory Werkspoor, we invited a select group of data enthusiasts from our clients to learn all about this booming specialization.

março 06, 2020

Hosting this Data Connect Session were Isabelle Liesker, our resident data consultant, and data director Paul van Oosterhout. In the 2.5 hours that followed, they passionately shared their extensive knowledge on the subject in an interactive workshop. The result? Six very useful insights (and a slightly fried brain).

The holy trinity of data at Valtech: understand, respond, predict

Before we get started on the insights, a small introduction on data at Valtech might come in handy. Being one of our 7 services, we’ve been working hard the past couple of years to show the relevance of and need for data to our clients. Our mantra is “understand, respond, predict”. You first need to understand your data (and your visitors) to be able to respond to, and eventually predict, their behavior. We’ve noticed a lot of different levels of maturity in data at our clients, depending on their company’s legacy – or lack thereof, in the case of start-ups. But no matter the data savviness of a company, understanding, responding and predicting are always key steps in your data roadmap.

Want to know more about data at Valtech? Get in touch with Paul or Isabelle!

Insight 1: Data is bigger than you think (and no, I’m not talking about big data)

Of course, data is a great way of getting to know your customers and understanding, responding to and predicting their behavior. If you’re not using data to its full potential, you’re missing out big time. Not only because you don’t know your audience, but also because data is a good (maybe even the best) starting point for breaking down silos in your organization. Because, as Paul said in the session, “everyone in your company is working on the customer journey”. They may not all be aware of it, but every department collects some form of data. Bringing all these data sources together will do magical things for your company.

                                               Everyone in your company is working on the customer journey

Paul van Oosterhout, data director

Insight 2: In 2020, we’ve got 4 data trends

Isabelle and Paul defined 4 trends ruling the data world this year.

  1. Automated data science
    As Isabelle put it: “How do I clean my data and do machine learning with it?” This trend announces a new level of maturity in data and opens the door to an endless list of possibilities.
  2. Data science in the cloud
    Locally stored data is so 2010’s. Both the growing size of data and importance of privacy mean the cloud is the way to go when storing and streaming data, and performing data science.
  3. Data privacy and security
    GDPR, AVG and CCPA. Do I need to say more?
  4. NLP (or Natural Language Processing)
    NLP is all about how to program computers to process, analyze and eventually generate natural language data. Though we do not use this with our clients yet, NLP has a lot of potential.
Insight 3: The biggest challenge of (big) data

It seems nowadays that people are obsessed with ‘big’ or ‘bigger’ data. It is exciting stuff, but if you really think about it, it’s really nothing new. As Paul puts it: “Big data is just data, but big. That means it also gives you bigger problems”. The biggest challenge for data in general, according to Isabelle and Paul, is the value aspect of it. How do you make actionable insights, based on a set of data? A very important question! Luckily, they had their answer prepared.

                             Big data is just data, but big. That means it also gives you bigger problems

Paul van Oosterhout, data director

Insight 4: 4 steps you need to take in order to gain useful insights from data

  1. Collect
    What are you going to measure and for what purpose? A rather important question, especially in the light of GDPR regulations.
  2. Ingest
    The filtering step. So, you’ve now got a shitload of data, but what part is actually usable? Pro tip from Isabelle: pick your data tools and let them do the work for you. 
  3. Clean
    We’ve determined which data is useful, so now it’s time to get rid of the useless stuff inside the data.
  4. Visualize
    Key in understanding (and making other people understand) data: visualize it! Graphs, box plots, pyramids, infographics – as long as it is beneficial to comprehensibility, the world is your oyster. 
Insight 5: Can’t see the (data) wood for the trees? Do a Data Design Studio

One way (and we would argue: the best way) to follow the steps mentioned above, is by starting out with a Data Design Studio. During our workshop in the Werkspoor, one thing became very clear. No matter the business you’re in, determining touchpoints and corresponding data sources is always the best starting point for data improvements. Generating use cases in a Data Design Studio enables you to break through the silos in your company and easily shows the benefits of using data.

Want to know more about what this mysterious sounding Data Design Studio is all about? Contact Isabelle or Paul and they’d love to talk you through it!

Insight 6: Example of visualizing data: keywords relating to 'Corona'

After loads of theory, it was time for an example. Using programming language Python in the online (cloud) environment of Google Colab, Isabelle visualized the rise and fall of keywords relating to ‘corona’. Thanks to the API’s from data figure generation Plotly Express (PX) and Google Trends, creating the visualizations was (or at least seemed) relatively easy to do. Let’s take a look at some of the possibilities of this way of visualizing data!


Keyword interest over 3 months
All three keywords have two lines, one for the worldwide keyword trend and one for the Dutch. You can clearly see we Dutchies were relatively late to the party when it comes to searching for ‘corona’.

Graph of interest in keywords related to ‘corona’
Lots to see, but one thing is clear: there’s no guessing when corona first hit the news.

Box plot of interest in related keywords
When it comes to this boxplot, the wider the box, the bigger the ups and downs in search volume for that specific keyword.

Interactive map of interest for related keywords
Have a look at the difference in interest between the first (measured the 12th of December, 2019) and second visualization (measured the 27th of February, 2020).


Want to learn more about data science at Valtech?

Get in touch with data director Paul and find out how we can help you and your business!
Contact us