Data Visualisation: questions you should ask before. What we’ve discovered working on the Map of Polish Composers.
TL;DR
Ockham’s razor is a designer’s most important tool. Data visualization has a value greater than the “wow effect” if and only if it allows you to show something better or explore more conveniently. The more non-standard your solution is, the sooner and the more carefully you need to test it.
→ http://mapofcomposers.pl/en/
Background
Our task was to design and implement a website collecting information about Polish composers of contemporary music.
The challenge was to reconcile various needs: the final product had to meet high aesthetic requirements, present information in a digestible way, and offer a navigation more interesting than an ordinary search by name.
The whole process took place in a space shaped by difficult baseline conditions:
- a ready data set prepared for years by outstanding specialists in the field of musicology,
- very limited time for design and implementation due to an already scheduled premiere and promotion during an important industry event,
- narrow (estimated at about 2,000 people), specific, unavailable for research and diversified in terms of their real needs target audience — festival programmers and musicologists from around the world.
We had little time to find an attractive form of visualizing existing data that would cater for various search scenarios and navigation around the data set.
The Sweet Dream of Users who “Discover Data”
Provided not only with encyclopedic data, but also information where composers belong to in terms of aesthetic trends and environments, and dates where individual works were created, we followed the path trodden by many a designer of interactive forms in the cultural sector: a search for a form of visualization so interesting that it makes users “involuntarily engaged” in extended interaction and independent “data set mining”.
Creative reflection on possible forms of presentation led us to the concept of a seven-dimensional space mapped on a two-dimensional screen using color, saturation and specific forms of movement.
But when we asked ourselves, what value it brings to our users, the unexpected answer was — none.
The idea of a seven-dimensional space where composers are presented simultaneously in terms of their main currents, stylistics used, historical background and geographical proximity sounds appealing, but creating it requires very precise data.
Even when the data is there, you still need to figure out how to communicate this content to users who know only two-dimensional charts. In addition, in times of attention economy, the user wants to see the value immediately — it needs to be clear enough to win with other incentives competing for their attention.
7 Most Important Questions You Should Ask Before
Therefore, the most important questions you need to ask when you start to design data visualization stem from questions about the value that it offers to your users in the context of their goals, about the best form of presentation, and about data quality.
- Who are your users? What mental model do they use? Answering these questions is crucial. Without it data is just raw pieces information stored in a database.
- Do we really need visualization? Why? This is usually the most important question, the answer to which is surprisingly often, “No”. Numbers and words are also forms of data visualization, and due to their high intuitiveness and extraordinary flexibility, they are often simply the best form.
- What specific value does this visualization bring? How many sentences do you need to explain the need for it? (Any number greater than one should raise your suspicions).
- Which data are crucial because in the mental model and language of the users they divide the set into subsets, and which are only complementary?
- What is the simplest way to present these key data? Is this form self-explanatory? Is there a simpler one?
- Do you need an interactive form or should you just highlight relevant information? Don’t make it interactive just because you can!
- Do we know everything about the data that will be visualized? Sometimes the whole concept can be invalidated by a small detail: insufficient accuracy, lack of a few data points, or a format that prevents simple algorithmic processing.
The Solution
Finally, we decided to implement two visualizations of the data set, both using the possibilities offered by the D3.js library.
The first is a traditional timeline which organizes the entire collection in an objective and informative way.
Time is the dimension that most ruthlessly regulates affiliation to epochs, currents, peer networks etc.
The second visualization method is based on language.
Each of the composers is tagged with words: names of schools, trends and phenomena. Words related to composers define a space of concepts in which all creators are inscribed in different ways.
We’ve created an algorithmic solution that draws a “constellation of the composers” based on their gravity towards various concepts, from the most general to the most specific. The interaction is mutual: the place occupied by a given concept in space is determined by the places occupied by composers associated with it. This algorithm generates a constantly tense graph in which the concepts occurring together in different contexts, and the artist penetrated by similar ideas gravitate to one another.
Innovations are risky. Always have a backup plan
All products should be tested early and carefully. Innovations — earlier and more carefully.
Navigation based on data visualization is an innovation based on a whole series of presuppositions regarding the importance of various data and the comprehensibility of their visual representation. Even one weak link in the design process can be decisive for the failure of the whole enterprise. The only way to avoid all reefs is to test the prototype step by step, which in turn dramatically increases lead time and the necessary resources.
Therefore, in critical areas, a contingency plan is a must: a mechanism that allows you to achieve the same effects in a traditional way.
The Map of Polish Composers is equipped with traditional indexes of names and terms. This way navigation is possible without using complex visualizations. When we moved from a 100-item test data set to a 250-item set, it turned out that this is already a serious burden on mobile devices, slowing down operation beyond acceptable standards. Visualization did not provide additional information and it hindered access to basic data — that’s why for mobile devices we’ve decided to use only traditional indexes and hypertext links between individual entities.
Profile Anatomy
Of course, data visualization is not only interactive forms. Information about each composer, presented on an encyclopedic card, was also carefully designed in terms of content (we decided to include interesting facts about the composer’s life to bring them closer to readers and make them more memorable) and form:
Research and conclusions
When we started our research, it quickly turned out that for our target audience content had absolute primacy over form!
Users were interested in the merits: the criteria for selecting composers, the presence of specific names in the collection, the way information was prepared, the substantive credibility of the descriptions, and the readability of the content.
What seemed most important during the implementation process and what at the same time was the most difficult in designing and developing — the navigation based on data visualization — was of secondary importance for the recipients. They appreciated the unusual form, interacted with interest, but the key value lay elsewhere for them.
“The respondents were almost completely focused on the content, criteria for selecting composers, and the texts. The innovative navigation was a tempting invitation to travel into the unknown, explore facts and connections in a way that no other service they use allows them to.” — Malwina Otto
That is why it is worth asking yourself the above seven questions and always remember that what is most interesting for a designer is often secondary for the user.
Digital Product Team
Kordian Klecha (linkedin) — product design + project management
Martyna Wędzicka-Obuchowicz (link) — graphic design
Damian Zawadzki (linkedin) — frontend+backend developement
Piotr Migdał (https://p.migdal.pl/) — data visualization in D3.js
Kama Czechowska (linkedin) — product design cooperation, photos
Malwina Otto (linkedin) —user research, case study translation
Zuzanna Wiechowska (linkedin) — research & benchmarking
