being clever with color

Quick announcement: I'll be doing just one more public storytelling with data workshop in 2015, which will be in Chicago in mid-October and still has a few open spots. Details and registration can be found here.

I've recently given a short presentation covering considerations when it comes to the use of color in communicating with data (earlier this week, for Chartio, and last week at the Seattle DataVis meet up). In it, I cover 7 brief lessons:

  1. Color grabs attention
  2. Color can signal where to look
  3. Color should be used sparingly
  4. Color can carry quantitative value
  5. Color carries tone and meaning
  6. Not everyone sees color
  7. Color should be used consistently

For each of these lessons, I talk about some specific strategies and look at a number of real-world examples. In order to bring this content to a wider audience, I recorded it, and you can view it here:

Leave a comment to let me know if you're interested in seeing more video content like this in the future.

The lessons covered here and much, much more can also be found in my forthcoming book, storytelling with data: a data visualization guide for business professionals, available for pre-order on Amazon today.

visualizing opportunity

When visualizing survey data, it's always seemed to me like showing where you're at is only part of the picture. There is important context that comes with where you could be. I had a workshop last week where we discussed this idea in the context of a specific example and I thought I'd continue that conversation here. What do you do when you want to show not only a summary metric, but also give a visual sense of the potential or opportunity?

The data that prompted this conversation was initially shown simply in a table. There were several survey items down the left hand side, with quarters over time listed along the top. Within each cell of the table, two values were listed - the average rating (with responses falling on a scale from 1 to 5) and the "top box" (the latter is a common measure in customer satisfaction or CSAT analysis, that combines those at the top end of the scale, typically those responding "Agree" and "Strongly Agree" or the responses corresponding to 4 and 5 for the given survey item). To illustrate, I'll show just one of the rows from that table (I've changed the details to protect confidentiality):

It's a clean, simple table. Still, we can improve upon it. I'll go through a few different views. First, if I were designing this table, there are few steps I'd want to take to declutter it: 

  • Remove the blue background in the top row of cells. Color can be used more strategically than this. In this particular example, I stayed away from color altogether, except in the heatmap that we'll look at momentarily.
  • Pull years at the top out into super categories and then have categories of quarter within that, eliminating the need to repeat years across the column headings.
  • Remove the sample size (N). If there isn't anything super interesting about it, this is typically more like footnote material, so let's imagine I have a footnote that says something like "Number of responses is relatively consistent over time, ranging from 350 to 472 over the quarters shown." (Related thought: in cases where the numbers swing by a lot, I would want to keep the N directly in the table so that can be taken into account when interpreting the numbers, but here they are all sufficiently big enough that I don't think they add much, so I can reduce the clutter they add to the table and capture in a footnote instead.)
  • Focus on a single metric. Multiple metrics tracking pretty much the same thing can become quickly confusing and unnecessarily complicate discussion. In this case, I chose to focus on the average, because I think it's easier to wrap my head around than the top box, but different scenarios would call for different decisions here. 

Here's what my resulting table looks like:

When I see a table, however, typically the first thing I want to do is try to visualize the numbers. We can start by doing that directly within the table, using conditional formatting to apply a heatmap where relative saturation gives us a visual sense of relative numeric values. Here's what that could look like:

I chose to draw the most attention to the negative end of the scale, making lower numbers a higher saturation of orange and higher numbers lower saturation of orange. Here, we see that Q2 2014 is markedly below BENCHMARK and that numbers go up (becoming less orange) over time as we move rightward across the table.

We can take it a step further and visualize the numbers in a graph. We're looking at data over time here, which means my brain goes first to a line graph. Here's what it could look like: 

I've preserved the same overall tabular structure when it comes to rows and columns (the original version had more survey items than this, so you can imagine additional rows, each with a line graph similar to this one plotting the respective data). I can now visually see the increase over time. I've implicitly assumed that the specific numbers are important enough to label directly (but also provided the axis on the left, so I can attach the labels EXCELLENT and POOR right to the numeric scale they've been turned into). As I look back over this, I realize throughout that I probably should have made it clear that the metric that's being plotted is the average across survey responses, so let's imagine I've done this.

Going back to my post title - one of the things I want to show in this case is the opportunity. Line graphs don't really provide that. Sure, we can see it's a 5-point scale and that the line is lower than that. We could even draw a line along the top (or extending out from the BENCHMARK point to highlight that we've improved but are still below it). Or, I can build in this idea of opportunity into every data point by visualizing the unachieved potential on top of our current data:

Personally, I like this last view the best. If I were only showing the "Overall rating" (as above), I'd probably want to make the bars a bit taller so you could really see what's happening over time. But in more of a dashboard view where you're looking at many survey items at a time (imagine row after row similar to this) the current view can work well for being able to see what the data looks like over time, where we are relative to benchmark, and so on. I like the use of the unfilled outline to give this visual sense of where we could be.

If it's of interest, you can download the Excel file with the above visuals here.

What do you think of this approach? What other scenarios have you encountered when it is useful to show potential or opportunity? What strategies have you employed? Leave a comment with your thoughts!

an unexpected Twitter frenzy + HelpMeViz

I routinely post to Twitter about data visualization and storytelling related stuff: interesting articles or blog posts, updates on what I'm working on, good and not-so-good visuals from the media, etc. A couple of weeks ago, I posted the following:

The full Economist article where this visual appeared can be viewed here.

The full Economist article where this visual appeared can be viewed here.

I honestly wasn't expecting anything close to the response I received - a couple dozen replies and quite varying views on the effectiveness of the visual. Some were fans; some were not. Others noted points of confusion. A number of people suggested alternative approaches (one vote for a dreaded pie!) some even taking a stab at what a remake could look like. There was a bit of dialog on having a venue for conversations like this so they can potentially be useful to others later:

On this last note, I'm going to parlay this post into a recommendation for a related resource. 

HelpMeViz is a site run by Jon Schwabish where individuals can submit visuals and ask for feedback. In the description, Jon says, "This site is designed to facilitate discussion, debate, and collaboration from the data visualization community." A quick scan through the homepage will reveal a lot of great and varied visualization examples, challenges, and conversations. In the event that you'd like to weigh in on the Economist waffle graph shown above, we've posted it on HelpMeViz here. In any case, I recommend checking out HelpMeViz and adding it to your data visualization resources.

how I storyboard

Storyboarding is one of the most important things you can do early on in the planning part of communicating with data to help ensure that your overall communication makes sense, meets the given need, and tells a story. A storyboard is a visual outline of the content you plan to create (ideally, before you actually spend time to creating content). I've written about it before here.

Rather than talk more about what storyboarding is and why it's important, in this post I'll share with you an example of some storyboarding I did this past year for my book and give you a sneak peek at one of the visuals that resulted from this process.

Chapter 2 of my forthcoming book focuses on choosing an effective visual display. In it, I cover different types of common displays of data and look at examples of and use cases for each. There are a plethora of different types of graphs and other visuals out there, and so I knew going in that I wouldn't be able to cover all of them. I went through several rounds of drawing different visuals I commonly use on Post-it notes and playing with the arrangement to come up with something that would make sense. But while I knew my approach wouldn't be entirely scientific, I realized as I was going through this process that it needed to be more scientific than that.

So I tried a more systematic approach. I went through every workshop presentation and all of my blog posts and consulting projects over the course of a year, categorized the various types of visuals I created, and tallied how many times I used each. I was surprised to learn thatout of the very many different visuals out therethere are exactly 12 that I find myself using over and over and over when it comes to communicating with data in a business setting. If interested, you can see the stats and relative rank ordering of my use of the various visuals and some related posts here.

Once I landed on the 12 types of visuals I on which I would focus, next came the question of how to organize these into a chapter in a way that would make sense to someone else. I turned back to my Post-it notes. I drew each visual on a Post-it and looked for similarities across them. I grouped them one way, then another way. I came up with categories. I got out a giant blank piece of paper and started arranging and rearranging my Post-its there. Eventually, I landed on a structure I was happy with and added some words and connections on the big paper directly to create my basic plan of attack for the chapter. Here's what it looked like:

The pink Post-its across the top are the main sections within the chapter. The blue Post-its show the subsections within each, along with the specific visuals I cover and some notes on other content to include. The eventual chapter ended up following this pretty closely. I did this sort of storyboarding to plan content for most of the chapters in the book.

My use of low tech pen-and-paper-planning didn't stop there. Next, I thought, wouldn't it be cool to kick this chapter off with a single visual showing all of the types of visuals I use most? I mocked up what this could look like on a blank piece of paper

My mock-up of a comprehensive "visuals I use most" visual.

My mock-up of a comprehensive "visuals I use most" visual.

After mocking it up on paper, I turned to my tools to create the final version. The following is a facing 2-page spread at the beginning of Chapter 2 in storytelling with data: a data visualization guide for business professionals.

For a related example, recently, I came across this post on Ann Emery's blog, which details a similar process that she went through when creating her Essentials Chart Choosing Tool (using index card instead of my favored Post-its). If you aren't familiar, be sure to check out Ann's awesome Excel tutorials.

Meta-lesson: starting low techwith simple tools like Post-it notes, pens, and papercan be a great way to get your thoughts in order and give you a plan of attack for creating your narrative and visuals and reducing time and iterations when you do finally turn to your more sophisticated tools.

and the winner is...

Earlier this month, I ran a visualization challenge, inviting improvements upon a few world population visuals published by The Economist. BIG THANKS to everyone who participated and spent their precious time creating and sharing their visualizations, which I'll in turn share with you in the following post.

One of the things I love about data visualization is that it sits at the intersection between science and art. When it comes to the science side: there are absolutely guidelines and best practices to follow. But there is also an artistic component. This means that two people faced with the same data visualization challenge may approach it in totally different ways. Or, as we'll illustrate here, eleven different people might approach the same challenge in different ways.  There's great room for diversity of perspective, which is awesome. That's one reason this space is so fun. 

Before I highlight the awesome submissions (with some commentary on each), let's look at The Economist's visuals that inspired the challenge in the first place:

See original with accompanying text here.

See original with accompanying text here.

The following submissions are shown in the order I received them. Make sure to scroll to the end, where I'll pick my personal fave!


Submission 1: Amber Smart

Amber was an early entry, which meant she had time to take the feedback I shared with her and incorporate it. Here's the first Tableau visual that she shared:

This is a sharp looking visual. I like how the regional breakdown is shown at the top right visual so that it's possible to see the trends over time relative to each other as well as have the summary stat of the overall growth in the circles at the far right.

The question I posted to Amber was on the choice of colors in the slopegraph at the bottom. I'd be inclined to have all of the lines the same color unless I wanted to draw attention to one or a few and make those different. Also, it's interesting how the placement of the #1, #2, #3, etc. in the bottom graph could cause you to read the graph slightly differently - the way it is now with the labels at the left, I make observations like "China is currently #1, will move to #2" whereas if those labels were instead on the right to help draw attention there first, instead I'd start with observations like "India will be #1 in 2050, whereas it is #2 today." It is interesting how relatively minor design decisions like this can change how someone reads and interprets.

Here is Amber's slightly modified visual based on my comments (nicely done!):


Submission #2: Mitesh Parekh

Mitesh also created his visualization using Tableau, commenting that he is new to the world of data viz and that he's testing his skills with this submission. I gave him early feedback on making sure everything is clearly titled/labeled, which he updated in the version shown below. I like how some of the graphs are titled with the respective question that they answer. I also like the takeaways outlined along the right side of the visual, giving the reader some meta-points to focus on. One point of feedback would be on the use of color throughout the visual, as it isn't always clear to me what the difference in colors represent (or whether they represent anything at all). My advice in general when it comes to the use of color is to not use it to make something colorful, but rather to focus attention where you want your audience to pay it. Overall, good job!

Submission #3: Leonard Murphy

Leonard created his visuals in Excel, using some templates that I've previously shared (for example, this and this). I promise, I won't let that bias me! He also mentioned that this is the first time he's tried some of these techniques (a great low-risk place to test them out!).

Here are Leonard's comments: I took cues from the text of the Economist's short article to determine what points to make. My first chart is almost identical to the Economist's, but without the colour-coded regions. I converted their 2nd set of charts to a single chart to make a clear point about how much Africa & Asia contribute to total growth. In my 3rd chart I wanted to show how much faster India is growing compared to China (in real terms). Sadly it has the effect of making Nigeria's growth look lackluster. I didn't want to split the slope charts into 2, for fear that the scale (and therefore slope angles) would change misleadingly. My final chart is as much to show the top 10 most populous countries in 2050 as it is to show their growth.

Throughout I focused on 2015 > 2050 only. For consistency and for relevance. The world was very different in 1950, and by 2100 the forecasts are as much fiction as prediction. In fairness to the Economist, I suspect they're not using the charts to tell a story, but rather to fit a large amount of information into as small a space as possible.

 Here are Leonard's visuals:

Leonard's original graphs were on two pages, so if anything is too small to read here, it's my fault, not his!

Leonard's original graphs were on two pages, so if anything is too small to read here, it's my fault, not his!

My take: the graphs are very clean and I like how Leonard not only showed the data, but also called out something interesting to focus on with the subtitles. As a point of feedback, if he wanted to take this a step further, he could make it even clearer why some of the data is highlighted (but not other) in blue by also making the subtitle text (or relevant portions of the subtitle text) the same blue. This would just be another visual connection showing that the text color and graph color are related. I also recommend adding axis titles (this is a point of feedback that is relevant for a number of the submissions - always title and label your axes!).

I like how the first graph is ordered by decreasing population. The final graph could benefit from this, too, which would make it more quickly clear that the countries expected to grow the most on a percentage basis are the ones that are highlighted (there should be logic in order!). Nice work!


Submission #4: Shobhana Thirumaran

Shobhana took a totally different approach to the population projections using a dot plot to illustrate the anticipated change by region (left graph below):

Hers was another early entry, and she iterated based on some feedback I shared with her. Overall, I think these visuals are nicely done. I found the graph on the left very interesting (I wouldn't have thought to show the data in this way). One minor adjustment would be to sort the geographic areas in order of decreasing size to help create a construct for the audience to use as they interpret the data. In the line graph, one approach would be to make all the overlapping countries at the bottom (and perhaps adding a single title for "Other" to make it clear that those are other countries shown for reference but not the focus). My final piece of feedback was to title the axes directly - you can figure out what each is from the graph title, but when you label directly your audience doesn't even have to think about it.

Here are the modified visuals from Shobhana (nicely done!):

Submission #5 - Geoffrey Felix

Geoffrey caught the blog post announcing the challenge when he returned from vacation, with less than a day left to create his visualization. He took a minimalist approach in Tableau, using lines and slopegraphs to summarize expected population growth in the various regions over time. Here are his visuals:

I like the callout in the first graph that points out when world population is expected to exceed 10 billion. I'd love to see similar callouts in the other graphs. I like how the change each region makes up of total is depicted via slopegraphs in the middle section and how he further details the absolute change in each area via the respective country breakouts at the bottom (I'd perhaps try to make this tie between the middle and final visuals even clearer). It's interesting to see that, in spite of anticipated huge numeric growth in India (final graph, left), the percent Asia makes up of total world population is expected to decrease (middle visual, left). 

Note how we're able to make totally different observations with this view of the data than was possible with some of the other visuals. That's part of the power of having multiple people visualize the data in different ways. Great job!


Submission #6 - Eliza Sienko

Eliza's was another case where she came across the challenge without much time remaining, but she put in quite a bit of work, looking not only at the data that informed The Economist's visuals directly, but also much of the additional data included in the full report. Her three pages of visuals follow:

When it comes to Eliza's visuals, I love that she's called out the story and key takeaways via the text on the right. I also like how she combined tabular and graphical elements in a single visual in the second and third visuals, allowing there to be a lot of detail while still leveraging visual processing via the bars (related blog post).

In terms of constructive feedback, I was a little confused on the use of color. In some cases, the colored text ties directly to the color used in the accompanying graph (for example, green Africa in the first graph and green text describing Africa's big contribution to the population growth), but that doesn't seem to always be the case (for example, in the penultimate graph, the population getting older in text and Africa in the graph are both green). Being consistent in the use of color (and not using the same colors for different things) could help make the visual ties between the text and data clearer for the audience (though with so many data points, that's not always easy!). Overall, nice work!


Submission #7 - Chris Saden

Chris worked with a combination of tools to create his visualizations, ultimately refining everything in Inkscape, turning them into SVGs that scale well (as opposed to PNG and JPG that can become pixelated). He commented that he thought the more interesting part of the story was the comparison from what happened in the past to the projections. He pulled additional data from 1950-2015 for the first to graphics to show that comparison and to highlight that 2015-2080 are projections. Here are Chris' visuals (which he's also written about on his blog here):

I think Chris' visualizations are beautiful. The story is called out above each graph. Color is used sparingly, consistently, and strategically to emphasize and de-emphasize (I really like the way the forecast figures are set apart visually from the historical actuals). Everything is labeled, so there's never any question what the data represents. The thoughtful design fades to the background so you can easily focus on the story and the data. Awesome job!

The things I would do differently here are minor and more personal preference than best practice. Chris, though, clearly a perfectionist, was pointing out issues he sees with his own visuals (a cut off label, perhaps too much white space, the final visual being a little pixelated). We tend to be our own worse critics!


Submission #8 - Jeff Harrison

Jeff led his submission with the following commentary: It took maybe two minutes of exploration to figure out that Africa was the big story in the data, and a skim of the UN report summary told me that it was a story worth telling. The main problem with the original wasn't a bad choice of graph; it was the decision to present a laundry list of population facts rather than pointing out what such a rapid increase might mean for billions of people. The presentation needed a "so what," as we say in my office. There are other stories that I could have chosen, of course, but this was the most compelling to me. 

Here is Jeff's visual:

This is another beautiful visual. I love the simple design, sparing and purposeful use of color, and the text that tells the story - any graph used to communicate data like this should have a "so what" as Jeff points out! I think it's clever how he titled the y-axes with the top data label - I haven't seen it done that way before. Jeff did a great job on the slopegraph, emphasizing just the increasing portions of the lines for the African countries but leaving the other data there for context. I'm not sure whether the country labels are needed both on the left and the right (could perhaps reduce to just on the right), but this is minor. The inclusion of the map is nice (and does well illustrating the point that not all of the growth is equal). I might have made the map shades of orange to make it clear it ties to the orange in the other graphs (and depending on how the countries in the slopegraph measure up, perhaps have them in the same shade of orange as each would be on the map - though this would be a bit more work!). Great job!


Submission #9 - Sanika Mokashi

Sanika created a full blog post detailing her approach to the challenge here, visualizing the data in the form of a Tableau story and focusing on a similar message as the original Economist article. She also found an earlier post from The Economist with the exact same visualizations based on predictions made by the UN two years ago. Here is Sanika's 3-panel Tableau story:

Overall, I think Sanika's visuals are beautiful. Everything is clearly titled. Key takeaways and stories are called out through a mix of subtitles and annotation directly on the graphs. I like the consistent use of color across each, with the map legend at the left (though I was still craving x-axis labels for the regions in the right hand side of the second set of visuals). I was a little uncertain on the inclusion of the far right graph in the first set of visuals (I think it's the same data that's plotted in the main graph in that section?). The separated waterfall chart in the second panel of graphs is an interesting approach for showing current population and also expected growth/decline by continent.

The images above are static, but in Sanika's blog post you can click the buttons at the top of the final visual to see the same graph for the top 10 countries in Europe and Africa as well (each with their own callouts of interesting things to note). Well done!


Submission #10 - Marie-Eve Vesel

Marie-Eve completed her visuals using Excel. Here are the comments she included with her submission: I decided to combine information from the first two original bar graphs into a single column graph.  In my opinion, it’s a waste of space to use a graph only to show the regional % change when that same information can be effectively visualized in a graph also showing the actual population from 2015 to 2100. In this case, I think columns are easier to read than bars, especially to compare the size of each region over time, and also to see each region’s contribution to the total column for each year.  I also emphasized the strong expected growth in Africa by using arrows larger than for the other regions (admittedly, I might have gone slightly overboard with the size…)

As for the most populous countries (on page 2), I opted for a slope chart split into three panels to avoid confusion from too many overlapping lines.  Each panel focuses on a few countries, based on their 1950 ranking. I included the vertical axis to make the reader aware of differing scales for each panel.  As opposed to drawing a straight line between the data points at 1950, 2015, 2050, and 2100, I used the actual underlying data to display the real population curve (yep, I’m an actuary and I like numbers, lots of numbers).  I also differentiated between actual and forecasted values by using a solid line for actual, and a dotted line for forecasted. Finally, I included the ranks at the 2015 and 2050 marks so the reader can see how each country is placing over time. While I think the original graph was hard to read, I liked that they color-coded each country according to its region; so I kept that feature in my revamped version.

I really like the line graphs on the second page. They are well designed - great use of color and visual cues to set apart the forecast figures from the historical data. Marie-Eve packed in a lot of info without it feeling overwhelming. The one thing I struggle with is the difference in scale across the three graphs - I can totally understand why she did this to be able to actually see the details in the smaller countries in the second and third line graphs. Normally, I advise against changing the scale on similar graphs because it can be confusing for the audience, but the way everything is labeled clearly here and also with the graphs placed below each other (vs. side by side where there would be greater temptation to compare the heights of the lines across the three graphs, which the reader shouldn't do) - I actually think this works. In other words, I think Marie-Eve changed my mind. Nice work!


Submission #11 - Ned Haugton

The final submission was from Ned Haugton, who used this as an exercise in "colour scheme generation." Here are his comments: The colours are generated by assigning hues to each group (rainbow with 80% saturation and 80% value), and then using a sobol sequence to generate scale pseudo-random perturbations (±5% hue, ±10% saturation, ±10% value) that ensure that each country has a colour that is unique, and also very distinct from it's neighbours (by population). The code is available here.

I think this is an interesting approach (and thanks, Ned, for sharing your code!). You can see the expected growth in world population and get a sense of relative breakdown across continents, with big growth in Africa's population and slowing growth in Asia. Other observations are more difficult due to the stacked nature of the graph. I do like how this is totally different than anyone else came up with. Interesting approach!

and the winner is...

So I sat here with all of these beautiful visualizations trying to pick my personal fave. It is not an easy task! 

When it comes to effective visualization, at a meta-level, I look for four things: 1) a sensible display, 2) absence of clutter, 3) affordance in design, and 4) a clear story (see related post with more detail). There were a number of submissions that I thought did a great job on all of these fronts.

Then I realized that this is my contest. Which means I can do whatever I want. More specifically, it means that I don't have to pick a single winner. Rather, I'm going to pick my favorite three that each do a great job of meeting the criteria outlined above.

The winners are (in no particular order):
Chris Saden | Jeff Harrison | Sanika Mokashi

Chris, Jeff, and Sanika, I'd like to invite you each to write a guest blog post for I'll reach out to you to discuss specifics. Really great work!

Big thanks to EVERYONE who participated in this challenge. I'm honored to be able to feature everyone's entries here. I hope you enjoyed the challenge as much as I did!

get the details right

First, announcements: There are still a few spots left in upcoming public workshops in NYC and LA: click here for details. Also, the DATA VIZ challenge is underway, with entries due 8/12, details here. Now back to our regularly scheduled programming.

Get the details right. It's a simple (and perhaps obvious) tip, but so very important. Let me recap a situation I recently witnessed when the details weren't right and the resulting repercussions.

I was observing a vendor presentation that preceded my workshop at a client offsite. I enjoy watching others present and considering what makes the good ones effective and what makes the not-so-great-ones not-so-great. In this case, two guys were presenting. There were things to learn from each.

The first was charismatic and easily held the audience's attention. What made him effective? He had a booming voice that he inflected to emphasize the important parts of what he said. His gestures were natural. He seemed at ease. His confident voiceover made me feel like he knew his stuff. It was when he passed the stage over to the second presenter that things turned south.

The first slide that presenter number two flipped to misspelled the client's name in the headline.

As if this weren't bad enough, the leader of the client group actually spoke up to point out the error. If anyone in the audience had their attention elsewhere and might have otherwise missed it, this ensured everyone was aware of the mistake.

The presenter was obviously embarrassed. Unfortunately, there was little he could to to regain the credibility this oversight lost him. Personally, my thought was, if he can't consistently spell the client's name correctly, what does that say about the attention to detail that he paid to the research he had done that he was working to convince them of? Fairly or unfairly, an audience might assume it doesn't mean good things. They may think this quietly. Or they may go on the offensive to try to determine whether the presenter knows his stuff. This group did the latter.

From this point on, the audience - who had asked only two questions during the prior presenter's 20 minutes of talking - peppered this second presenter with inquiries. Very detailed questions. One after another. They picked him apart. It wasn't pretty.

A couple lessons can be learned from this. First, definitely spell check anything you're going to put in front of your audience, particularly if it's going to be written in 40-point font. And the meta-point of this post: details matter. Get them right.

This was a particularly egregious error (especially with the leader pointing it out). More commonly, when it comes to data visualization, the issues I encounter looking at client work are numeric. For example, percents that don't sum to 100% when they should, numbers that should be exactly the same in multiple places in the same deck (or on the same slide or graph!) being different, or math that just doesn't work out the way it should. If you show math, make sure it's correct. All it takes is one discerning member of your audience to realize the numbers don't add up, which can call into question your credibility in the same way that unfortunate presenter number two had happen to him. Use spell check. Double check your math. Have a friend or colleague review your work for a second set of eyes to catch things you might miss.

Whether a full presentation or a single data visualization: don't undercut your credibility with your audience with silly mistakes. Details matter. Make sure you get them right!