A friend of mine recently asked about my process for designing dashboards, namely “requirements gathering.” When left to my own devices, my process lacks fancy acronyms and is designed around the unpredictability of humans. Let’s face it: humans are predictably unpredictable. In this process, the dashboard becomes secondary. Scary thing to say as a consultant whose job IS making Tableau dashboards.
Let’s look at this, though…what’s the real point of these things anyway?! To me, they’re communication pieces, vessels for having a conversation with data. You see, I’m fairly fluent in “data-speak” or VizQL or what have you. Numbers don’t scare me…in fact, I like to poke them much like I do weird food (most of what I eat). However, it seems I’m fairly alone in this Sheldonlike love affair of numbers and logic.
When I first started this process, I had at best an hour to design something, usually for my boss. It usually went down like this:
Boss: hey, I need a dashboard.
Me: sweet! What are you thinking?
Boss: I saw this big spike on our demand dashboard and want to look into it. Where’s it coming from? Any idea of what’s going on?
Me: sure! Let me get started on this. When do you need this?
Boss: I have a meeting in 30 minutes and this will probably come up.
Me: on it!
I’d then bust into the data, facet it a dozen different ways until I got some better questions, and then hit the interwebs. Chances were high with random spikes, it came down to weather or newsworthy events. Sometimes, I’d pull down other data and mash it up with what we were seeing and the story would be clear. Other times, it’d simply be random, and a few times, I could point back to a few other reports and show where Nostradamus the forecast predicted this.
I started to formalize this process a bit more after taking a course on design thinking, or human centered design and when I got days to design a dashboard, not minutes. It’s the same process IDEO uses and it’s the method used to create the iPhone. Love it or hate it, it changed the game on the smart phone.
So, what is this process? It’s 4 simple questions that drive each phase and a few steps in each part. Every part has something visual and tangible about it. At the bare minimum, there’s stickies involved. How can you go wrong with sticky notes?!
Design Thinking: the Magic at Work
Design thinking, or human-centered design, starts out with 4 phases. I give all props to Darden for this.
This is the part no one likes. It’s a rehash of everything currently in play, competing factors (such as Excel reports), and an understanding of the players. It involves a ton of sticky notes and what the experts call “user empathy,” or what I tell folks is my 101 crash course on doing the work.
Let’s face it: I may think the world of Tableau and consider it the answer to life, the universe, and everything, but most of my users tell me they have real jobs and work to do. I get it: I used to be in their shoes and understand that they may only have minutes to glean insights from this dashboard. So, it’s crucial I know as much as possible about real human beings. I’ll even go as far as mapping their journey: what does a “regular” day look like (I hear you snickering over there).
Most people try to tell me that “everyone” will use this dashboard. I usually point them to server if I can, so they can see it’s mostly Sheldon that will be hitting this report, or ask for the pickiest person (usually Sheldon). Sheldon gives me the inside in: the fact that weather plays a role, that certain days mean things will happen, or this unrelated thing (fresh donuts at the Krispy Kreme) somehow affects demand in some way that even the Force can’t explain. These things matter.
Once I get inside Sheldon’s head, I trace his steps through the chain. I poke the data, ask more questions, and start to scare people with my psychic data sensemaking abilities. Wanna creep people out? Analyze their data and tell them they’re a Sagittarius. Make sure to validate this on LinkedIn.
This is the part people want to do first. Or, they want to start at step 3 and come back to this. This is that brainstorming part where spaghetti hits the wall. People love it, especially when it’s not their wall. Or it’s shellacked with silver paint. It’s a thing I guess.
Here’s what gets people though when we do this: we simply throw ideas out there. We don’t judge what’s possible or what’s not, what’s expensive or too simple, or any of these other filters. No, we just toss spaghetti out there. If it’s going to stick, it will. Usually, this involves stickies in some fashion and everyone has a chance.
After we’ve come up with every crazy idea, then we filter. We often take some of the outlandish ideas, scale them back a bit, but use them. People are always amazed by this. No, they get used to the filter or that if it doesn’t stick right off the bat, then it’s a dud. But here’s the thing: sometimes, that strand will stick if it’s shorter or given a bit more time. These are usually the ideas that win big in executive meetings.
This is that magic moment when the airplane lifts off the ground, when childlike wonder takes hold, and you realize that you’re really flying. It’s awe-inspiring and creepy all at the same time. I’m usually the punk staring out the window, watching the city become smaller and smaller. 5 year olds are predictable and some of us never grow up.
People begin to realize about this stage what Tableau dashboards can do. Now, some want to jump right to this stage, but you miss out on the real nuggets you find in the first two stages. Yes, I can certainly do a line chart here, a bar, some mapping, and a table for drill down, but it may miss the real opportunity.
By waiting, we get a clearer picture, an understanding of the intricacies that really drive business. These may not be what we want or even expect, but finding them paints a far better picture of the world. Who would’ve predicted that I’d want a phone to read my fingerprint…it seems Big Brotherish to contemplate what I would’ve thought 10 years ago, but now almost all my devices can do this.
The goal here is prototyping, not a completed product. In fact, I expect changes. Sometimes, I miss the mark completely or we realize what seemed good on paper doesn’t cut mustard with the data. Fast results means quicker answers and plenty of room to turn around. The more we have these conversations at the start, the better the results are at the end. When doing a larger project (usually not Tableau), I’ll go as low resolution as possible. Tableau makes it quite easy to deliver something quite elegant quickly.
We go through a couple iterations here until we hit the magic point.
This is when we go live or wide scale. Usually, by this point, we have clarity and have a rock solid dashboard or multiples of. I’ll finish up any loose parts on the style guide and work through the next series much faster. This process works anywhere where humans are involved and solutions need to support them.
That’s my process. It can be as long or as short and as formal or informal as it needs to be. I love it and think it works well Tableau.