One of the key things that interests me as a non-data scientist is how best to present data in a way that is both appealing and easily usable.
It's a commonly-accepted message that a picture is worth a thousand words, but too often I've been presented with a graph (that's kind of a picture, right?) and expected to understand everything there is to know about a complex piece of information in just a few seconds and with minimal supporting text to help me.
You wouldn't hand someone a thousand word essay in a meeting and - pausing just long enough to sip a latte - say, "So, you can see why we need this $10,000 project to be up-and-running by Thursday."
But in real life this kind of thing happens metaphorically every day.
I read on a website recently about data scientists and designers working together on a sleep app (https://hbr.org/2018/03/what-happens-when-data-scientists-and-designers-work-together) and how they seemed to be trying to address this challenge.
I think the article I read was making a point about having a person-centred rather than data-centred approach – and that considering design is a key factor in that.
I agree. Data scientists need to work together with both the end users of their data - and those who commissioned it, the copywriters, artists and software manufacturers who decide how that data is presented – as well, of course, as the designers, in order to make their hard work and potential insights best serve the people who need it.
So how should data be presented? Is it just about design? Is information delivery always “data + design = message”. I suppose it depends how broadly you define design.
As I've already said, I'm not a data scientist. Or a designer. So there are probably many factors I'm unaware of. But I am a data user. And I know when I've had something explained to me properly.
I want facts and opinions, with the two clearly demarcated. I want all the information to make a competent decision on the thing I’m concerned with…and unless I ask for it, nothing more. But if I want more, I want to be able to dig down and get it easily. I want everything to be up-to-date, and if it changes I want to know when that happens and why.
I’d like things to look “trustworthy” too. No unfinished sentences, mis-aligned tables, or coffee-stains. It may be illogical, but my gut needs to feel happy.
I don’t think I’m unusual in this.
If you are a service supplier presenting data in a mobile phone app for a customer who wants information quickly to make a low-value choice - on the go - about what to do, you probably would do well to present information differently from if you are a product supplier presenting information for a retailer sat at a workstation deciding whether to make a high-value decision at their own pace.
For example, if I am using a mobile broadband package on my smartphone with a 100mb free and then £1 a megabyte thereafter, I would probably rather receive an automatic data shut-down at 100mb and an icon to tap to continue using mobile data anyway when I reach 99mb, rather than a complicated (or even simple, for that matter) graph detailing how much data I've used, what I use it for most of the time, when I use it most, and what my predicted costs will be if I carry on as I am doing.
But if I'm a soft drink retailer with a regular supplier then I would much rather that company sent me graphs showing trends, buying habits, my projected costs and so on, rather than a simple message that I can afford to order one million cans of pop, would I like to go ahead because it's probably going to be hot and sunny next week. I want greater insight.
In both cases I want clear – but differently presented – information…but is it more than a design issue? Is it about different teams working together? Or, if the data scientist is given a clear brief as to what the goal is to start with, would the outcome be the same? Better?
Some might argue that in information delivery it is as important to have clearly-defined and narrow-enough goals before starting to mine data as it is about presenting the data that we end up with from our efforts. Some might argue that goes against the whole philosophy of the discipline.
What’s for sure is that for me it's not enough to present data accurately. It needs to be presented in a tailored and timely way as well.
But that probably comes with a price tag, so I look forward to many more latte-choking meetings in the near future.
What do you think?