The Power of a Single Question

Just over a month ago now I had the privilege of speaking at the Fanalytics session at the Tableau Customer Conference 2013 in Washington DC. My talk, based on my own learning experience of Tableau and Tableau Public, centered around a key rule I found when developing dashboards...'Only answer one question per dashboard'. If your topic is complicated or the data is deep and interesting then make it multiple dashboards and walk the user through them. But as an end user coming in fresh to this particular representation of data it's difficult enough trying to understand the answer from multiple charts let alone remember which chart is answering which question.

Golden Rule

This simple rule, which I still hold myself to, came about by throwing myself into the deep end of data visualisation in the public space...my first publication on Tableau Public and The Guardian Datastore. Looking back at that first piece I was so excited to get out there, I can see I started off well. My reason for the work was to answer the question 'Was the UK in 2011 as dry as in 1976?' and as you can see that statement of intent is proudly displayed at the top. Great start!Then things started to go wrong. You see over the past 2 years, since Tableau got the full focus of my professional career, I've come to the conclusion that there are two types of dashboards. The first is the one question dashboard and the second is the explorer dashboard. Choosing the right one can mean the difference enlightenment and confusion.Unlike answering one question, which assumes the potential end user has a limited understanding of the subject matter and/or underlying data, the explorer dashboard is a tool for experts of both the subject and the data. An explorer dashboard doesn't set out to answer one particular question but instead frees a large or complicated dataset into something easily queried in many different ways via a data driven visual tool such as Tableau. But it will only be meaningful to a user who already knows enough about the subject area to develop their own questions.

Choose wisely or forever confuse

Turning back to that doomed dashboard, which received comments such as 'complicated', 'difficult to understand' and 'the moderator has removed this comment as the language was too strong to publish', it's clear to see that mixing these two types or picking the wrong one can spell total disaster. For the rainfall dashboard I'd set out to address a simple question but failed provide the simple answer, instead publishing an over complicated waste of my time. I'd waved the question in front of the user's face suggesting 'the answer's in there somewhere!' Result: confusion, frustration, bad language!

Putting the theory to practice

As a Tableau consultant I'd say my work is split 50/50 between these two types of dashboard. Sometimes a client has a collection of data that, before Tableau came along, was considered too big and messy to try and understand something from. The approach here, much like introducing the general public to a piece of interesting BI, is to help answer some basic initial questions...one per dashboard. For example 'what is the value of a repeat customer?', 'do people engage in a meaningful with our newsletters & social media channels?' and 'what is our overall gross profit percentage each week?' The rest of my work is based on a realisation that I will only ever understand about 10% of what the person sitting next to me does about the company/data, and instead I'm there to visualise the data in the best possible way so they can do their awesome work of making complex decisions even faster than before.As this is Tableau Design Month I thought I'd conclude with some advice to people starting out or hoping to develop their internal Tableau and Tableau Public work. Keep to the single question dashboard! It's not that the other type is 'only for experts', it's just focusing on one question is a tried and tested method. Tableau provides us with an awesome range of functionality that we can utilise to visualise data and in no way does this mean we can't use them. I'm not advocating boring dashboards which don't utilise actions and drill-down. But if before each move you make you ask yourself 'is this helping my audience answer the fundamental question?' you'll be sure to enlighten rather than confuse.Enjoy Tableau Design Month! I know I am, there's some awesome stuff being produced.