Data Changes Everything

Understanding visualization design handoff.

Photo by Glenn Carstens-Peters on Unsplash.

Complex data visualization projects often involve collaboration between people with specialized skills. Visualization project teams often include designers, developers, data experts, and project coordinators where each member of the team has skills geared towards specific phases of the project. Handoff is the exchange of information between the different team members with different roles working on a project; it is how the output of each development phase is encoded and passed on within the team.

A visualization design is influenced and inspired by the data it is relaying. Data is its core element and for this reason, data has the most impact, not only on the final design but each stage’s artifacts. Let’s look briefly at the different stages of a complex visualization project:

• Project Conceptualization: This phase refers to the decision to carry out a project. It occurs on the client’s side and is before the development team’s involvement. The artifacts generated from this phase are the project goals, visions, and the data set to work with. It is important to note that this data set is rarely ever complete but often contains enough data to show the project’s focus. Handoff at this stage is typically simple and occurs via email; it can also be a more complex handoff such as meetings or workshops with the client’s data experts.
• Data Characterisation: At this stage, the design team explores the dataset provided, prioritizing it’s characteristic concerning the project’s vision and goal, which is referred to as Data wrangling. This phase also includes understanding the dataset’s values, types, and relation with the project goals. The artifact from this phase is a summary of data characteristics, which is presented to the client for confirmation and kickoff. Insights from this stage are the foundation of the entire project.
• Visualization Design: At this stage data is abstracted and encoded into a visual design, it is also referred to as the Data Mapping phase. Here, a visual concept is developed and approved by the client; then the design is refined, polished, documented, and passed on to both the client and the development team.
• Visualization Development: This phase is led by the software development team, where they implement the designs handed off to them. The designer’s role at this phase is more recreational; and they participate by answering questions, suggesting redesigns where issues arise, and confirming the implementation works as expected.
• Deployment and Use: At this phase, the visualization is deployed. The development team is tasked with maintenance and periodic data updates. The design team gets involved if updates contain values the design does not support.

Now that we understand the different phases and the handoffs of each phase, we’ll look at the key ways data changes things and can complicate handoff.

Adapting to Data Changes

Data updates often have cascading effects. Updates can be particularly impactful when it changes the data characterization the design was built on. Even where data change does not affect the overall mapping, it might affect the implementation in terms of server-side mechanisms for loading, aggregating, and computing the data.

Anticipating Edge Cases

it is difficult and near impossible for designers to anticipate and test all possible combinations of interactive inputs that visualization might generate. So it is difficult to anticipate the presence of values that can break the design or data mapping.

Evidence of edge case data exceeding it’s design allocation

Understanding Technical Constraints

During the data mapping phase, designers are rarely concerned or aware of the constraints that govern the development phase. These constraints can be directly related to design like charts and graph combinations that are difficult to achieve or indirect constraints to design implementations like cross-browser compatibility, frameworks that must be used, code maintainability to name a few. Designers try to mitigate the uncertainty by prototyping and testing novel designs in code.

Articulating Data-Dependent Interactions

When ideating design interactions, designers often have to generate a variety of views to properly pass across interaction capabilities. The need to do this and the consequence of not doing directly translates to extra costs.

Communicating Data Mapping

Implementing data mapping often requires more detail and precision than is available in visualization mockup. The designer of a mapping must be able to convey their intent to others on the team, especially the developer. Difficulty in conveying intent can be a result of data complexity, interaction complexity such as multi-dimensional and interactive visualizations.

Preserving Data Mapping Integrity

Given that there are multiple phases (and consequently multiple outputs) in visualization projects, there is the opportunity for misinterpretation and misapplication at every phase. Differences can emerge as the development progresses for reasons ranging from bugs to data updates, design inconsistencies, and misinterpretation of mapping.

Misinterpretations in Visualization Design implementation.

The primary method of testing adherence to the designer’s intent is by visual inspection and comparison to the original design. With every iteration introducing numerous small changes and consequently opportunity for discrepancies, it is tedious to keep track of what has not been inspected. More so, visual inspection alone might not reveal all the issues.

Data is hardly ever defined during the implementation of a visualization project. And where it is defined, there is the opportunity for data updates to break the design. However, there are several opportunities for mitigating the complications of the impact of data on a visualization by utilizing:

• Data characterization tools: can help remove the stress of adapting data updates to designs by highlighting the differences in data sets.
• Data ideation tools: This would ease ideating data-driven designs. These provide better handoff opportunities from designers to developers.
• Data mapping frameworks: developing standard tools and methodology for communicating data mappings. These would support explicit communication of the relationships between data structures and their graphical representations.

This blog post is based on the following paper:

• Walny, J., Frisson, C., West, M., Kosminsky, D., Knudsen, S., Carpendale, S., & Willett, W. (2019). Data changes everything: Challenges and opportunities in data visualization design handoff. IEEE Transactions on Visualization and Computer Graphics, 26(1), 12–22.