Perspective: How a small dataviz tweak can tell a radically different data story

What do you see?

What do you see?

Is there a single best visualization for a given set of data?  Should you always use a bar chart when looking at categorical data with a single metric?  Is a timeline always best for viewing trends over time? Is there only one data story to tell?

There are a couple of different viewpoints on this issue.  Some would argue, “Absolutely there is one best way to visualize certain sets of data based on the question being asked! And it can be scientifically determined.” (See Stephen Few’s recent post and discussion in the comments).  Another group says, “It depends! There are many different ways of seeing the same data and what works depends on lots of different factors.” (See Cole Nussbaumer Knaflic’s post)

I lean toward the “there’s multiple ways to visualize data” viewpoint – but the debate not what this post is about.  Instead, let me offer the following observations and then move to the main point:

  • I think there’s more agreement between the camps than appears at first glance.
  • While I’m not convinced that there’s always a single “best” way to visualize data, there are often wrong (misleading, false) ways and “not-as-good” (confusing, lacking clarity, boring) ways that should be avoided.
  • What I love about Tableau is that it makes it easy to iterate through various views of the data – all the while asking, answering, and raising new questions – that lead to new visualizations that answer and raise new questions. The flexibility and ease of changing a visualization allows me to have a conversation with the data.  A single view gives me a single viewpoint.  Multiple different views of the same data allow me to see things I might have otherwise missed.
  • It is amazing the difference a single change to a data visualization can make – sometimes it changes the whole story – or at least gives a totally new perspective.

And that leads me to my main point: even a small tweak to your dataviz can uncover new perspectives and new data stories.

An example: Absolute versus Relative Time

Take for example, this view:


How many cumulative views of posts on

Tweak tells different data stories.

Notice the variations:

  1. X-Axis is Date: You see a running total of page views for every post on the popular Tableau blog,, starting with the date of the first view.
  2. X-Axis is Age: You see a running total of page views for every post, starting at day 0.


Now consider the similarities:

  • Both are timelines (technically, you might consider the second a scatter-plot).
  • Both have the same number of lines
  • Both have the same exact shape of lines


The only thing that’s really changed is the origin of the lines.

  • In the first, the origin is absolute: the date on which the post was introduced
  • In the second, the origin is relative: all posts start on day 0.

And yet, the data story that is told is very different.


Data Story 1: How Many and When?

Data Story 1: How Many and When?

This story is focused on “when” and “how many” and there are certain aspects of the story that are relatively easy to see:

  • When were posts published? (A)
  • How long between posts? (A)
  • What are the most popular posts (in rank order)? (B)
  • How popular were posts? (B)
  • How fast did posts grow in popularity? (C)


But contrast that with this story:

Data Story 2: How Fast?

Data Story 2: How Fast?

This story is focused on “how fast” and there are different aspects of the story that are easy to see:

  • How old is a given post? (A)
  • How popular has a given post become? (B)
  • How fast did a post grow in popularity? (C)
  • How fast did a post grow in popularity compared to others? (C)
  • How popular was a post X days after being published (D)


And now consider one final tweak that enhances the story in other ways:

Animated Data Story: Posts on

Notice how the animation adds a few elements to the story:

  • The absolute date becomes easier to see and understand again
  • There is a hint of relationship between posts (notice how once the most popular post comes into play, the trajectory of other posts change).


Now What?

Now you’ve seen my example. What examples do you have?  What existing data visualizations do you have that could tell a different story if they were tweaked?  What answers can you uncover in the data based on different visualizations?

So go home, take out your old data vizzes and experiment!  Let me know what you uncover!

And of course, you’d like to know how to build the relative time series and even animate it, wouldn’t you? You’d even like me to give you a Tableau workbook?  Absolutely! That’s the next post… (so check back soon!)


Best Regards,
Joshua Milligan

19 Responses so far.

  1. Great post! Nice animations!

    Changing boundary conditions will lead to a different optimum.

  2. Davi says:

    This post is fantastic.

  3. Sean says:

    How did you make that last animation gif?

  4. Sri says:

    Great Post ! Looking forward to the next post about the animation its definitely gives a very different perspective

  5. I have a silly question: what formula did you use to compute Age of Post?


    • Joshua Milligan says:


      It’s: DATEDIFF(‘day’, [Date], TODAY())

      The [Page] field needs to be in the view to define the level of detail — so what you are getting is the difference between a given date (of the page view/visit) and today. Hope that helps!

      • All right, that really helps, very simple indeed, I thought you were using some fancy table calcs involving a LOOKUP()… Keep it simple…
        I tried to insert a parameter in your formula to select other time slices than Day, such as Week, but no go, Datediff wants a STRING literal… Any idea what I am missing?

      • Another Separate question: How do you get the curve to stop on your running sum of pageviews instead of staying flat for the subsequent days without a value?

        • Olga says:

          Same here, I’d like to know how to keep the aggregate intact when adding date to the Pages shelf.
          And thank you for the post Joshua!

        • Joshua Milligan says:

          Mine stays flat. You could use a different mark type (such as circle or bar) which would let you see the gaps more clearly. The other alternative is to turn on data densification to pad out the domain of all dates and then format the measure to break the lines. I wasn’t able to accomplish that with my view in a minute or two – so I’d have to give it some more thought and effort to determine how. I’ll let you know if something comes to mind. (Or maybe someone else has a thought…?)

  6. Tony says:

    I need some help with dates! Not sure where to post this So I am trying here.

    What I would like to do is allow the user to Select either “Fiscal Year” or “Calendar Year” as a parameter filter and place “Order Date” as a column and have it adjust accordingly. The business wants to switch between Fiscal & Calendar for the dashboard(s)

    Is that possible?

    The alternative

    • Joshua Milligan says:

      It should be possible. Tableau allows for setting of a fiscal date. Assuming the fiscal year is relatively simple and just starts on the first of a different month, you can right click a date field and select Default Properties > Fiscal Date Start.

      That gets a single field. To swap between fiscal dates and actual dates, you might consider 1) using a sheet swapping technique to swap between one view using one date field and a different view using the other, or 2) using a DATEDIFF calculation to “shift” the dates based on the parameter selection.

      Hope that helps!

Leave a Reply

Your email address will not be published. Required fields are marked *