Have you been asked to do data visualisation work for an organisation or project, only to find that the they are unprepared for you? The analysis isn't complete, the audiences aren't well understood or even the messages they want to communicate have not been identified.
This happens a lot, so to prepare myself I've started to compile a checklist, which I either complete myself, or actually give to the client for them to self-assess.
None of the items in the checklist are showstoppers - but by being aware of these items early means you can provide a more honest and accurate assessments of the work to be done. There may be tasks you can help with, like understanding audience needs or infrastructure requirements, which you can add to your scope of work.
1. Is the data ready to be visualised?
- Is the data in a stable state? Are changes expected and how often?
- Has the cleaning, processing and analysis complete?
- Has the data been approved for release?
Too often an engagement will start, but then delays occur immediately as the analysis goes through peer-review and management sign-off, and you can't access the actual data to begin design and development.
Often waiting until the first round reviews is complete, or management have approved the release of data, will save wasted time early on.
If the nature of the visualisation will include dynamic data (dashboards or monitoring news events), knowing how often these updates occur (weekly, daily, minute-by-minute) can help you understand when it will be best to commence work,
2. What is the nature of the data?
- What are the dimensions of the data – tables, fields, types?
- What is size of the data?
- Where is the data stored and how will it be accessed by the solution?
The best thing is if you can see the actual dataset (also refer to section 1 above). If you can't, then at least understanding the complexity and volume will help you understand what performance and infrastructure considerations need to be factored in.
3. Are the audience’s needs well understood?
- What will the audience use the visualisation for?
- What is their familiarity with the subject matter and datasets?
- What interactivity is required for the visualisation (eg filtering, searching, linking with updates on other visualisations)
Another gotcha I see a lot: an organisation wants to create some funky, cool visualisations, but have no idea exactly who it will be aimed at. Understanding this is critical for any visualisation to be successful.
4. What technology constraints are in place?
- How will the audience view the visualisation – online only or print too? For online, is mobile first, desktop first, or responsive?
- What technology and platforms are available or restricted to?
Designing interactive responsive visualisations is hard - but worth it for some audiences (see section 3 above). The act of translating designs from print to online takes time, as not all elements work well in some media.
5. Is the design of the visualisation agreed or suggested?
- What charts or graphs are expected to be used, if any?
- What style or branding guidelines are in place?
Some organisations have great brand and style guidelines - even data visualisation guides. These can be a time-saver, but also a constraint if they prevent you from using colour palettes that are optimal for displaying certain datasets (eg sequential and diverging colour schemes).
And occasionally I've run into stubborn Communications teams that refuse to budge on certain colour choices, despite their ineffectual use for displaying data (eg the two primary colours are too close in lightness and saturation).
6. What are the stories or messages they want to emphasise?
- What the top 3 to 5 points that the visualisation should focus on?
I usually ask the organisation to write down the messages they want to emphasise in the visualisation. If they can't do this, or list too many, then you are making something that a) isn't going to succeed, or b) is going to be so full of options/filters/buttons that you may as well give the audience Excel with a pivot table and get them to explore the data themselves.