The Four Tableaus and other question responses

At the end of last week, I received some extra questions from the attendees who viewed my coffee break webinar. I’ve responded to each individually but thought I would share the responses with you as well.

But before I get to the questions (and responses) I wanted to share a little about the presentation experience. Feel free to skip to the next header if you don’t like to hear about emotions and such…


Handling Vulnerability cartoon by Chris Lysy

I don’t really get all that anxious before posting a blog post. But with live presentations, either on the web or in person, I am a wreck. I get all quiet and just stumble through my slides in my head over and over again. I never feel good about a presentation before it starts, no matter how well I’ve prepared.

When presenting live, and the presentation is going well (engagement through eye contact, questions, and maybe a few laughs), I tend to warm up and feel better by the end.

When presenting online, things change. Sitting alone in my office, talking to slides, and seeing/hearing absolutely no response (the chat was monitored by AEA) is an unsettling feeling. Most of the time when presenting on the web, feedback is delayed, and this talk was no exception.

After it was over I received a couple questions and a few nice comments. But not enough to make me feel better. I let my insecurities get the best of me. The people with the good comments were clearly just overly nice people, you know the kind that would never say anything bad.

Then flash forward a week and a half, when I get the feedback back from AEA…

I had 164 attendees, of which 57% (~94) completed a feedback form. It was a global audience with participants from Bolivia, Canada, Ghana, India, Netherlands, Spain, United Kingdom, United States, and Australia. 38% of those who responded were first time AEA webinar attendees.

“Your speaker rating was 4.67. On average, Coffee Break speakers rate a 4.32.”

“Your overall rating of demonstration was 4.49. On average, our webinars rate a 4.24. Average conference sessions rate a 3.7.”


  • This was SO helpful! Just a huge thank you!
  • Well-developed presentation. Need more like it.
  • Informative, yet very comprehensive
  • I enjoyed this. Please have him back.
  • Great demonstration
  • All around perfect performance!
  • I look forward to showing my team the video once uploaded
  • Definitely conference material
  • I want to thank AEA and the education team for putting on such an informative and well-presented presentation.

I felt really insecure about something that was, according to the feedback, a really solid demonstration. If others feel the same way about online presentation, it’s no wonder that only a fraction of people who present live even consider presenting on the web.

Now on to the questions!

Inexpensive Tools

Do you have tools you would recommend that you mentioned might be inexpensive?

This all depends on your personal context.

If you have the capacity to code, javascript libraries are really inexpensive. If you don’t have that capacity, they are not. Most people I know in evaluation do not have that capacity.

If you can publish your data publicly, Tableau Public is really inexpensive. If it’s not ready for public, but someday will be ready, start with a dummy dataset that matches your dataset structure. Put together a working prototype, then when the data is ready, refresh and update.

If you can’t publish your data publicly, the cheapest way to use something like Tableau is $1,000 for a single license for one year. Every following year to maintain the license costs $200. You can still publish interactives and share privately, but the person receiving them must have a copy of the free Tableau Reader.

Qlik and Microstrategy both have free analytics programs. They hope to get you hooked so that you’ll purchase one of their bigger licensed products at a later date. I’ve tried both but I tend to use Tableau for most things.

Expensive cartoon by Chris Lysy

To Code or Not to Code

Do you ever use tools like high charts, fusion charts, or zing charts? They are not the same I suppose but seem useful.

To code or not to code.

There is a lot you can do if you know how to code. If you can, tools like the ones mentioned (let’s throw d3 in there too) can be incredibly useful.

If you want to take this approach, and don’t know how to code, you’ll need the help of a web developer. They need to be able to build a site (lots of ways to do this), code the necessary javascript (these sites give lots of example code and walk through the process), and work with data (databases/APIs).

Coding is free cartoon by Chris Lysy

The Four Tableaus

Tableau is an interesting platform that we have explored, however, I am concerned about ownership of the data. Do you have any other platform recommendations where the data remains fully owned and controlled by the user? Or, is my understanding of Tableau flawed?

There are essentially four major Tableaus (or 6 really, but I’m going to talk about 4).

Tableau 1: Public version where all data is stored in the cloud and available publicly to everyone.

Tableau 2: Desktop version where none of the data is stored on the web and is fully controlled by you. It can stay private and be opened/interacted with by non-licensed clients/colleagues using a free reader version. Or you could eventually upload to Tableau #1, #3, or #4.

Tableau 3: Online Version. Requires Tableau #2 in conjunction. All data is stored in the cloud but available only to you and whoever else you say. To interact with it via the web, the client/user needs an online user license.

Tableau 4: Server Version. Requires Tableau #2 in conjunction. This Tableau lives on your server fully under your control. You buy licences for everyone who will interact with it. This is where Tableau can get really expensive. Starts around 12K and goes up to several hundred thousand dollars.

Inconvenience cartoon by Chris Lysy

Controlling the Message

How do you stay in control of the message when letting people play with the data and see their own stories?

First off, if you really need more control over the message then an interactive will likely not be your best choice. Infographics and animations give much more in the way of control as you can structure the narrative as you see fit.

With that said, you can still do a good bit to guide your readers. Here are a few ways.

Color. If you have something specific you want the reader to follow (i.e. focus on a specific state/city/project) you can set a highlight color just for that something. Then mute the colors for everything else by default.

Filters. What you filter, what order you place the filters, and how prominent you make the filters is under your control. If you want your reader to filter by something specific then put that filter right up at the top. Maybe even number your filters or add text to guide your reader on what they should do next.

Tab Order. I’m not big on the whole> “this needs to be one page” talk that comes with dashboards. Your readers have the ability to click and Tableau has made it easier with their 8.2 update with new page navigation. Order your pages/tabs by what you want your reader to see, and what you want them to see next.

Tailored. Like any other piece of content, you don’t have to create only one view. You can tailor it to specific audiences and send them individualized links. Start them where you want to start them.

Data design cartoon by Chris Lysy

What to Present

How do I know what parts of my data people will want to interact with? Or do I just put it all out there?

Think hard about the questions your audience will want to ask. Use your evaluator skills and talk to the audience beforehand. Interactives, in the way I’ve presented, are first and foremost a question answering tool.

Say you’re working with school data and you usually report SEA (State Education Agency) means. Talk to the states and you might hear that they want the numbers broken down by LEA (Local Education Agency). They also might give you insight on some of the other important variables they would like to see alongside that breakdown. These would become the filters.

Also, use your instincts. What potentially interesting data do you have? What could help provide new insight?

You can also prioritize what you want to include. The major stuff is used to create the overall structure of the visual (i.e. your scatterplot’s x and y). The important things you want to use to break down the data make up your filters/colors. Then the stuff that is really only context for specific individual data points, stick that in the tooltip.

You will still be selective, but try not to go overboard on being selective. Give them what they need to answer their questions.

Complicated chart cartoon by Chris Lysy

Capacity and Training

What kind of training or eval capacity does this require among stakeholders?

Anytime you create a tool, even if it’s simple and seems really easy and intuitive, you will still probably need at least a little training. How much depends greatly on the complexity of the tool, the importance of the tool, and the specific audience.

Focus on pointing out how the tool meets an audience’s specific needs. How it answers their questions. The basics to how you explore the tool. And then how to quickly refresh and start over.

It’s easy to go overboard on showing how the clicks/selects/filters work but you have to prove the benefits first. If you can really show how this tool meets a specific need, your audience will work through the learning curve. If you don’t, or they can’t see it themselves, it doesn’t matter how good your instructions/training matierials are, it won’t be used.

Training cartoon by Chris Lysy

Visible Data Issues

What are issues around sites having their data made visible to others?

There can be all sorts of issues. Some things to keep an eye on.

Personally Identifiable Information. If it would be a problem to publish the underlying dataset, don’t publish the interactive. Also, just because someone else collected it (i.e. Google Analytics) does not mean it’s not filled with PII.

Sensitivities with data sharing. Some project sites might have an issue with their data being shared with other project sites. What is appropriate for your situation is always going to depend on context.

Overly cautious. The easiest thing for any of us to do is never share anything. Presenting data should never be taken lightly, but don’t dismiss by default. Talk to stakeholders, think about the implications, and think about the potential value.

Interactive visualization tools can also be very useful for analysts and project management. Using the tools does not mean you have to make things public.

PII cartoon by Chris Lysy