Launch, grow, and unlock your career in data

April 20, 2012

Nielsen’s Advertising & Audiences Report – A poor infographic redesigned for simplicity


Nielsen published this infographic as part of their Advertising & Audiences Report:

Ok, Nielsen, I see that you’re trying to be cute. 

  1. You made an old fashioned TV the background image (which many young people won’t even recognize)
  2. You used blocks to represents a parts-to-whole relationship.  Maybe you used blocks so that children would be reminded of Legos.
  3. You chose background colors to remind us of how a TV would look when it had no signal way back when

But it’s all a bit too much for me.  I find myself trying to count each of the blocks.  My eyes are darting back and forth to the color legend.  The value labels sometimes don’t even line up with the associated color.

Are you getting consulting from David McCandless?  You must be! 

Simplify things.  Communicate more effectively.  Save yourself time.


This took all of 10 minutes in Tableau.  How long did it take you to create your infographic?  I bet significantly longer.


  1. You are correct there are many issues with the original, and yours is a vast improvement.

    With the understanding that no visualization is perfect, and that your view was made very quickly, I would like to offer some options that could improve what you created.

    - display the axis
    - have a meaningful sort
    - remove the legend
    - switching hierarchy

    With these changes applied, I came up with the following:

    Share of Viewership

    I believe this view makes it easier to spot the outlyers and see the overall trends.

  2. Thanks for the feedback Joe.

    I agree that I should have removed the legend. However, I don't think the scales are necessary because you're trying to measure the relative sizes of the bars.

    This article from Stephen Few helps explain the reasoning. Go to his example at the end.

    The sort is in alphabetical order by genre, the same as Nielsen. I did that intentionally since that's the sort Nielsen had and I was trying to make a direct comparison.

    It doesn't make sense to change the hierarchy because the categories are not related. The comparison should be genres within each category.

    I thought about sorting them by rank within each category, but then I thought it would be confusing not to see the genres in the same order in each section.

  3. I have to say, despite the issues of comparing blocks vs. bars, I really prefer the original. There is value in having graphics that look a bit unique. I often find myself blazing right past yet-another-line-chart or yet-another-bar-chart, but will stop and pause on something that looks different. Without knowing the goals of the Nielsen blog, you could speculate that if they succeeded in getting people to pause for a moment on their graphic, the graphic was a success.

    This is a simple data set, and with all the #'s printed right there, I don't have a problem with the blocks. I was able to easily learn some interesting things, like there was no spending on product placements on sports shows, and could confirm some things that make sense (people don't time shift news shows, probably because news gets stale pretty quickly).

  4. This comment has been removed by the author.

  5. Thank you Andy, I have taken your comments into consideration and made two more views:
    - Alt 1
    - Alt 2

    My current preference is for alt 2.

    I feel it is easy to see that Drama gets a sizable chunk of the share across all categories, while Reality has a moderately low percent in most except for a very high percentage in product placements. I also found it interesting that Sports had a low percent of time shifted, and that news has a low share in all categories.

    I prefer a single axis over the 2x2 matrix to make it easy to compare the lengths of the bars.

    I asked Stephen Few for his thoughts on the removing quantitative scale in a 2x2 matrix, and he said: "I would usually include the scale. This is because I would want to make it obvious that the scales on all graphs are consistent."

  6. Shawn, I like your's the best. It's the same as Joe's #2, but easier to read since it's bars not columns.

    Joe, thanks for following up with Stephen. His article had only one chart, so having four in mine, I can see where the axis would help. I initially had them.

    Stephanie, I suppose if people are drawn to the viz because of its flashiness, then they achieved their goal. However, all of the flash and the little blocks add no value to the information they are trying to communicate. In the end, shouldn't the post be judged on the quality of its content and not it's ability to snatch your eye?

  7. This is a good representation of ratings on the television, May I know what stock charting tool you are using? Thanks!

  8. Nice post and very neat improvement proposals from you all!
    I completely agree with all your comments and concerns on the Nielsen's visualizations.
    Nevertheless, I would like to add that the visual metaphor used here is not that bad. Although the proposed barcharts are better, the metaphor of the waffle chart is also a good alternative when we do not have that many values. The interesting value of waffle charts is that they help the reader to see that all the parts sum up to 100%, which is something difficult to see in the barchart approaches.
    IMHO, the worst thing in this chart is the color palette along with the disturbing background images.

  9. Right on target Andy. Much improved, fast, easy!

  10. Anybody for some bubble maddness?

    Great stuff - awesome original critique Andy and I loved all the angles people took in the comment section.


  11. I like it Ben! I still prefer the bars, but the bubbles are a nice alternative too.

    Like you, I have really enjoyed the discussion that has been sparked.