1. Skip to navigation
  2. Skip to content
  3. Skip to sidebar

Our Blog

Talk to _people_ about your data

Ken Hilburn

I need your help. I want to ask my wife if she’d like to go see a movie with me. Which way do you think I should go:

Option 1:

“Interested in catching the 7PM showing of Hunger Games at the AMC?”

Option 2:

Time: 7PM

Movie: Hunger Games

Location: AMC

Attend?

Well, we both know the answer: she might be inclined to take me up on my offer either way, but with the second option, she might also be inclined to make me sit on a different row.

We know that when we talk to people, we have to do so in a way that feels warm and personal. Why is it then, that when we design systems for people, we want to woo them with option 2?

Imagine a sales report where you select a date range, the type of transaction you are interested in and the resulting metrics. It might look something like this:

Start date: 1/15/2012

End date: 4/2/2012

Type of transaction: Closed deals

Revenue: $1,500,000.

Number of deals: 6

On the other hand, why don’t we treat these sort of data interactions more personable; more like a conversation? Take that same report and re-imagine it in the metaphor of a sentence. Something perhaps like this:

Between 1/15/2012 and today, there have been 6 deals that closed, totaling $1.5m in revenue.

Nifty? We think so. To see some examples and learn more, check out this latest addition to our design principles library.

Here’s to treating people like… well, um… people!

Related posts
  1. Dashboards talk too much. Most dashboards just talk and don’t listen. Have you...
Topics:
  • Kevin Phair

    It’s nice in theory but in many cases when we need information we’re looking for something specific in it. In your example, if I was just interested in the revenue number I could look at the table and almost instantly find the revenue line. When the data is converted to conversational form I have to parse the syntax of the sentence to isolate the elements one by one until I find what I want (unless I happened to remember that revenue was the last item mentioned).

    Obviously if I wanted to isolate the revenue line I could ask your report generator specifically for it, but it does mean having to submit a new request. I would envisage this being a good solution where you have a lot of back-end computing power so that a request is processed and results given almost instantly (a bit like gmail; You don’t miss not having folders because it’s so quick just to type in what you want to find if it’s not categorised by one of your top-ranked labels).

    It’s something that would be very open to abuse by well-meaning UI/report generator developers where they think they’re making things easy but haven’t thought it all the way through to the end-use case (as so often happens).

    One of the things I used to like about using the Amiga OS is that you could choose to have dates formatted in a human manner, so if the date was within a week it would give you the name of the day, or even “Yesterday”, “Today” or “Tomorrow”.

  • khilburn

    Kevin,

    You’re absolutely right in that if someone is looking for a specific number and understands the context of the result, columnar data is definitely the way to go. In that case, the context of putting the result into a more descriptive sentence could be counter productive.

    If you haven’t seen it yet, we actually have a short video on the proper way to make data more scannable and more easy to read in a table format on our Resources page: http://www.juiceanalytics.com/white-papers-guides-and-more/ (look for the Videos section and the “Table Design Best Practices” link.)

    You’re also correct that we need to carefully navigate the slippery slope between theory and practice. Thanks for reminding us of the difference between theory and practice: In theory, there is no difference. In practice there is. ;-)

  • http://twitter.com/victorblaer Victor Blaer

    Good points.
    Additionally find that the answer is regime specific. If people have never heard of data viz I find it helps in the change management process to use the conversational style and a table. It makes them feel secure and comfy. Then as they get used to the dashboard (Tableau in my case) I can start taking the training wheels off.
    This is so much easier than to force feed/quote the data viz lords ;)