The Evolving Role of Data Visualization: 3 Key Takeaways for Practitioners

On June 6th, 2025, the VizKnowledge conference took place at Aalto University, close to Helsinki, Finland. As two engineers from the Data Visualization squad, we were keen on attending this conference, which this year was held under the banner of “Resilience”. It aimed to show through the media of data visualizations “how individuals, societies and systems adapt to a rapidly-changing world, and how their connections stretch, break, and are created anew”. It is always exciting to see how other practitioners from a variety of backgrounds are using data visualization and information design to explore current topics, and we were not disappointed.

02-cafe-regatta.png

We also had the chance to explore a bit the beautiful city of Helsinki, learned a lot about the Moomins (the cute trolls living in the Moominvalley), had some fights with the hungry seagulls that tried to eat our Karjalanpiirakka (Karelian Finnish Pie with Egg Butter) and enjoyed the hospitality and kindness of the Finnish people.

In the following section, we will recap our favorite highlights of the conference.

04-helsinki-food.png

Menna El-Assady: Intelligence Augmentation

One of the standout presentations at the conference explored the fascinating field of intelligence augmentation, specifically how to bridge human and artificial intelligence. Menna El-Assady showed several examples of how visualizing the process of an LLM can help create models that are tailored to the user and improve the model’s reasoning through expert input. The talk emphasized co-adaptive analysis, where the model teaches the user and vice versa, and the innovative use of multiple competing models for user satisfaction.

06-menna-el-assady.png

El-Assady highlighted two key applications: first, using network visualizations of political debate data to allow users to analyze topics and trace arguments in near real-time. Within minutes after the debate, she was able to get key argumentative structures visualized only based on the transcript. Their models were also able to find the most convincing arguments from Reddit threads based on which arguments people interacted with most.

Another example that was quite fun was the common issue of bias, where if given free range, the model would turn all women into blonde housekeepers and receptionists and men into football players, presidents, and guitarists. When trying to debias this model, the result was an almost completely genderless world where pronouns couldn’t be sorted correctly to common names. El-Assady made the interesting argument that truth is not a technical problem. So, because we live in a biased world, we would also need to allow a certain amount of bias in a model, depending on what use case we want to show.

This is where the co-analysis comes into play. Given an expert to train models by comparing and rating their decisions, in order to get the best model for their field, allows for better results. Using the AI agent's search capabilities and computational power while also utilizing the advanced powers of humans for semantic connections and perception can combine for a powerful spectrum of intelligence.

In a world where AI becomes predominant in our lives (especially as software engineers), finding innovative ways to tailor it to our needs is key.

Watch Menna’s talk on Intelligence Augmentation here

Ben Fry: A Faster Way to Understand Data

Next was the session with Ben Fry, who is not only the founder of Fathom, a data and design studio, but one of the creators of Processing, the tool with which I (Laura) learned to code. So it was a huge highlight for me to see him in person.

Ben shared insights from his team's extensive data visualization work on pandemic response. His team went on a long journey of building adaptable data systems to tackle real-world problems, from tracking COVID-19 cases in universities to deploying real-time health data in a dashboard in a matter of hours. His message was clear and powerful: data visualization shouldn't be the final, polished output of research, but rather an active, exploratory tool used throughout the entire process. By embracing messy, incomplete data from varied sources, visualization can become a critical part of a rapid and dynamic response, rather than a historical summary published months after a phenomenon.

07-ben-fry.png

It was particularly interesting for us to see the new tool born from this philosophy, called Rowboat. It answers the challenge many data practitioners face: the time-consuming initial exploration of a dataset. Rowboat was designed to transform a raw spreadsheet into an interactive dashboard in seconds, automatically generating distributions, value counts, and simple charts without a single line of code. He called Excel a bad storyteller and Rowboat a great narrator that instantly shows you the shape of your data. This is something that we could think of integrating in Celonis products, helping users get an instant analysis of their data and showing key metrics so that they can start understanding and shaping their analysis faster.

You can try the tool here or as a browser extension here.

08-rowboat-tool.png

Watch Ben's talk on a A Faster Way to Understand Data here

Angela Morelli: Co-Design is a Necessity, Not an Option

One of the most thought-provoking sessions came from Angela Morelli of InfoDesign Lab, who argued that for data visualization to be truly effective, co-design is a necessity, not an afterthought. Especially in fields like science and climate change. She defines co-design as a collaborative process where designers and stakeholders work together from the start, making visualization an integral part of the scientific workflow that shapes content, rather than being a decorative afterthought.

She also shared that effective communication is not just about clear facts. It requires understanding the audience's values, social context, and emotional connection to the data. Providing more data or clearer charts can fail because it ignores the human element. The co-design process is the best chance to create communication that builds bridges, avoids polarization, and resonates on a human level.

09-angela-morelli-quote.png

An example of all of this was her work with the Intergovernmental Panel on Climate Change (IPCC). An intense process of creating visualizations that had to be approved word-for-word by hundreds of international delegates. One critical choice born from her co-design strategy was the decision to replace a traditional map with a cartogram, ensuring small island nations were visible and central to the climate change narrative. Without a co-design process, these islands would have been overlooked because of the geographical size, making invisible the real struggles and impacts of climate change that its population faces.

10-angela-morelli-ipcc.png

For us, this is interesting in how we can bring different perspectives together from our customers and stakeholders, to build products that truly resonate with them, fix their problems, and create value for their businesses.

Watch Angella’s talk on Co-Design here

Juuso Koponen: Explaining climate change and biodiversity loss

After winning a prize for their work in design and journalism, Juuso Koponen and his creative partner Jonatan Hildén got the opportunity to create an exhibition on a topic of their choice - the critical issues of climate change and biodiversity loss. Koponen, in his talk at the VizKnowledge conference, emphasized that while climate change is relatively well-understood, with iconic visualizations like Ed Hawkins' warming stripes, the challenge lies in connecting individual actions to the broader environmental impact.

11-warming-stripes.png

Warming stripes graphic published by climatologist Ed Hawkins. The progression from blue (cooler) to red (warmer) stripes portrays global warming from 1850 (left side of graphic) through the date of the graphic (right side).

Additionally, biodiversity loss is often less understood and sometimes even denied, with ecosystems appearing healthy on the surface while being hollowed out from within. Measuring biodiversity loss is complex due to its multifaceted nature, but data from bird populations and insect biomass show alarming declines.

12-juuso-koponen.png

Koponen showcased experimental approaches to explain these issues, including minimalistic maps highlighting biodiversity hotspots and infographics illustrating the environmental impact of urban sprawl versus dense building. He also tackled the controversial issue of forestry in Finland, explaining how burning wood for energy, despite being politically deemed carbon neutral, negatively impacts forest biodiversity and carbon targets. Carrying with him a medium-sized stone with a handle, Koponen provided a tangible visualization to represent per capita emissions and the forest's diminishing carbon absorption capacity, urging for sustainable forest management and increased protected areas.

13-juuso-koponen-islands.png

Using humor and creativity, Juuso Koponen exemplified how to use the power of visualization to support a cause that is, or at least should be, of utmost importance to all of us.

Watch Juuso’s talk on climate change and biodiversity loss here

Our takeaways

Visualization as a process: Instead of seeing visualization as a refined end result, during the conference it was framed as an iterative, collaborative process. This means we can use data visualization as an exploratory tool from the very beginning either to understand data or the needs of different users and stakeholders.

Visualization as a bridge for AI understanding: In a world where the use of AI tools is spreading widely, visualization has great potential for visualizing the internal processes of LLMs, allowing people to understand, interact with, and train models. Where more and more processes will start to rely on AI for its inner workings, creating visualization tools for users to tailor LLMs to their specific needs is a great opportunity for product innovation.

Visualization to shape understanding in unexpected ways: Data visualization is not just a tool for presenting data. It can also be a way to fundamentally shape our understanding of it. For example, representing the German legal code as a circular network tree as Corinna Coupette did in her Visualizing Legal Complexity talk, imposes a new structure on the information. This design choice doesn't just show the data but it can also change someone's mental mode, shifting their understanding from a simple list of rules to a complex, interconnected system.

14-german-code.png

We were very happy to join the conference and learn from some of the newest research and projects in the data visualization field, and are sure this will enrich our future work at Celonis!

Laura Junco

Laura is a Software Engineer in the Data Visualization Team. With a degree in Design, she aims to bridge the gap between the technical and visual aspects of data visualization. Laura and her teammates specialize in creating best-in-class visualization tools for all of Celonis Products.

Donna Klamma

Based in Munich, Donna joined Celonis as a Software Engineer in 2022. She is passionate about collaborating to create a fun and functional data visualization product. Together with her team, she aims to build a tool which delivers a more joyful experience than the typical enterprise software.

author

Donna Klamma