Welcome to the deep dive. We sip through the noise to find the signal, the key insights, and today you're right here with us as we dive into effective graphic design principles.
That's right, we're drawing from Leroy Bessler's Visual Data Insights using SaaS ODS graphics.
Now it is focused on SaaS software.
But honestly, the core design ideas they apply pretty much everywhere, very.
Universal stuff totally. So our mission today is to pull out those really practical, actionable tips. How can you make your visuals clear, more impactful.
And Bestler kicks things off with well of the fundamentals. He boils it down to three key objectives for any visual Okay, what are they Providing precise numbers, showing what's actually important, and crucially making sure it's readable. Clarity.
Right. The precise numbers part is interesting. It's like the visual gives the overview, but the data underneath. Yeah, that's the foundation exactly.
And he brings in that great quote from Joseph Pulitzer. Oh yeah, be blief, clear, picturesque and accurate.
That's fantastic advice and it connects directly to what you are listener are probably looking for getting the knowledge quickly, you know, without feeling overwhelmed.
Absolutely brevity, which leads perfectly into the next big idea focus. Bestlaer shares this great piece of advice from his assistant, which wise, just put it on one page.
Huh direct, very direct, Yeah.
But powerful, right, limit the message, don't overload people.
And he has that anecdote about Miller brewing the computer report.
Oh yeah, the discapacity report. Apparently it was just pages and pages long, kind of useless until until he's upseted it. Focus just on the key yessage categories. Suddenly boom, the important stuff just jumped out.
It's a perfect example, less clutter, more impact.
But and this is important, Bestler stresses, you have to be transparent when you subset, meaning meaning you need to tell the viewer what's included, but also what's not included. Give them the context.
Okay, so how do you do that?
He suggests, using titles and subtitles clearly state you know, the toward number of categories versus how many you're showing, and give the grand total versus the subtotal for what's displayed, maybe even the percentage of the total.
Ah. I see, So like top twenty cities make up seventy two percent of total sales. Something like that.
Exactly like that. Yeah, it gives the viewer the full picture even while you're focusing their attention. No surprises, got it.
It's about guiding, not hiding. Okay, so let's shift gears a bit. This other category. Bessler has a fun name for one approach.
He does, the pac Man pie chart.
Okay, explain the pac Man pie chart.
Well, think about it. Sometimes you have a bunch of really small categories. If you group them into one tiny other slice, yeah, it visually emphasizes how insignificant they are compared to the main players. Like pac Man about to eat a tiny dot.
Okay, I get the visual. So it downplays the small stuck.
Right, But then the opposite can also be true. A large other slice can actually peak curiosity. Bessler uses an example from his local government property taxes. They showed the village's slices small, and a big chunk was just labeled other.
Making people ask exactly, Okay.
Where's the rest of my money going. Then it prompted questions about that big other chunk.
Huh. So the other slice isn't just a dumping grounds. It can be a communication tool.
Itself precisely depends on what you want to achieve. Minimize something or maybe highlight a collective you want people to think about.
Interesting. Okay, moving on sparse graphs. This sounds like it connects to the keep it simple idea.
It absolutely does. Simple graphs are just easier and faster to grasp. Less cognitive load for the viewer makes sense.
Any specific examples.
He mentions his sparse line annotation for trend data, basically stripping away anything that isn't essential to seeing the trend itself.
So minimal gridlines, maybe fewer labels if the trend is obvious, something like that.
And he points out that those pac Man pies we just talked about they also count as sparse because they're so focused and simple.
God keep it clean. Now, something really fundamental. Bar charts, specifically the axis.
Ah, yes, the starting point. Bethler is very very clear on this, and the rule is for bar charts without negative values, the value axis must start at zero. Period.
Why is that so critical?
Because if you start at higher, say just to zoom in on the tops of the bars, Yeah, you completely distort the visual comparison. A bar that's twice as tall should represent twice the value starting above zero rates that fundamental relationship. It's misleading even if you don't mean it.
To be right. It exaggerates the differences visually, like that five foot tall growth chart idea.
Exactly zero baseline unless you're dealing with negatives. It's about visual accuracy.
Okay, And this ties into another principle he calls maximally simple design.
Yep. Again, it's about removing clutter. But here the distinction is important. It's not about limiting the information, but.
About limiting the unnecessary graphic.
You got it, extra borders, weird background patterns, three D effects, anything that doesn't actually add information or clarify the data.
Get rid of it, Marie Condo for charts you said earlier.
Kind of does it spark understanding? If not, maybe it doesn't belong.
Okay. Now let's talk footnotes. Often tiny text at the bottom right, Usually yes.
But Bessler argues that if the footnote contains really crucial information, like a major caveat or definition.
Which wouldn't be tiny, right, he actually.
Suggests making it the same size as the title text, maybe even bold.
It wow, really that prominent?
If it's critical context? Yes, it signals. Hey, pay attention to this. It fundamentally affects how you read this chart. Boilerplate stuff, fine print, sure, keep it small, but important. Context needs visibility.
That makes a lot of sense. Okay, what about the text within the shart itself? Access labels? Values?
Big point here? Use horizontal text whenever humanly possible.
Because it's just easier to read, way easier.
Our eyes scan horizontally naturally vertical text, tilted text, it forces you to crane your neck, slows you down.
So what about, say, long labels on a vertical axis.
He suggests trying to integrate that information elsewhere. Put in the subtitle, maybe the main title if.
It fits naturally, or sometimes you don't even need the label exactly.
If the horizontal axis is clearly showing dates, do you really need to label it date? Probably not, The context makes it obvious.
Good point. And what if you have lots of tick marks on the horizontal access and the labels start overlapping or turning vertical?
Yeah, that's a common problem. He suggests a few things. Can the viewer reasonably infer the missing values? Maybe you only label every second or third tick.
Mark, or strength the font shrink, the.
Font carefully, or some software has a stagger option putting labels on two levels. Anything is better than unreadable vertical text.
Okay, good practical tips there. Now, thinking about graphs online on the web, any specific advice definitely?
First, mouseover data tips are pretty much.
Essential, so when you hover the mouse, you.
See the exact value for that point or bar. It adds that layer of precision without cluttering the main visual.
Nice. What else for webgraphs?
For really dense data thinks scatter plots with tons of points or complex trend lines, consider linking to the underlying data, maybe an Excel file.
So people can dig deeper if they want to.
Exactly gives them the option for detailed inspection without overwhelming the initial graphic. You can link it right from the graph page.
Smart and sizing, oh yeah.
Size the image appropriately for the screen. Show them the entire picture, as Bessler puts it, avoid forcing people to scroll around just to see the whole.
Charge right, basic usability. Okay, let's tackle the infamous pie chart people love them or hate them? Bestler's take.
He gives best practices. Key thing include the slice description, the actual value and the percentage for each slice.
Where do those go on the slice in a legend.
Either way, but make sure it's clear. He mentions using call out lines if needed to avoid labels overlapping, especially on smaller.
Slices, and the order of slices.
Definitely order them largest slice to smallest slice, usually starting at the twelve o'clock position and going clockwise. It guides the eye to what's most important.
Makes sense. What about that other slice and pies?
Generally, he advises against using another slice in a standard pie chart unless you're specifically doing that pac Man thing we talked about earlier for a very deliberate communication purpose. Otherwise it can obscure too much.
Okay, got it. Color obviously huge in visuals. What are the key color considerations?
Contrast is king, especially label text against its background slice or bar color. Dark backgrounds need light text, Light backgrounds need dark text, simple but critical.
Any tricks to make that easier?
He suggests choosing palettes that are either all generally light fills or all generally dark fills. That way, you can usually stick to one label color, either black or white for everything. Simplifies things.
Interesting idea. What about the use of color itself, when to use it when not?
He has this great line. Uncolor might be the right color.
I mean black and white basically, Yeah.
Black and white if you have very few categories. Maybe shades of gray if you have a handful, reserve vibrant color for when you really need it to differentiate many distinct categories. Don't use color just for decoration.
So purpose driven color exactly.
He also mentions using transparency, which can be handy for overlapping elements like in scatterplots.
Good tip. Any warnings about color.
Oh yeah, the big one is how colors look different on screens versus projectors. Big difference.
Sometimes I've seen that happen washes everything out totally.
So test on the actual projector if you can. Even different computer monitors can show colors surprisingly differently.
So what's the safest bet.
Black and white is universally safe. Shades of gray are usually fine too, as long as they're distinct enough. Color always carries that extra risk of looking wrong on different devices.
He mentions SaaS HLS colors briefly.
Yeah, just noting that if you use shades of the same hue, make sure the lightness and saturation vary enough to be distinguishable.
Okay, And bringing it back to basics. Again, just making sure people can read the text.
Absolutely fundamental, sufficient font size, high contrast, plain background. Maybe use bold strategically for emphasis readability. Trump's almost everything.
Else makes sense if they can't read it, what's the point?
Precisely?
Okay, couple of specific chart types to finish on three D pie.
Charts just no, He's very blunt. They are always misleading because the perspective disorts the slice proportions ods graphics actually prevents you from making them directly.
Good riddance maybe probably, yeah.
Though older sends tools good, but yeah, avoid three D pies.
What about donut charts?
A donut chart is basically just a pie chart with a whole It uses the same logic, same options.
Then you put text in the hole.
You can. There are options like whole label and whole value, but he warns they have limitations. You don't have much control over the font size, and software might cut off the text or abbreviate it unexpectedly, so not.
Super reliable for critical info.
Probably not. He suggests using subtitles, footnotes, or maybe an inset statement if you want more control over text that appears in that central area, rather than relying solely on the whole labels.
Okay, and don't it charts still need ordered slices and clear label Oh.
Yeah, all the same best practices as pie charts apply. Order the slices, include category names, values, percentages, keep it clear.
Got it? So, wrapping this all up, lots of practical advice here.
Definitely. It boes down to clarity, focus, accuracy, and readability.
Prioritize the data, guide the viewer's eye, keep it clean, make sure everything's legible.
Use elements like subsetting or even that other category thoughtfully. It's all about creating those aha moments for your audience, not overwhelming.
Them right, helping you, the listener, get those insights across effectively.
Which leads us to a final thought for you to chew on.
Okay, let's hear it.
Think about a visual you're working on right now, or maybe one you made recently. Pick just one principle we've talked about today, maybe starting bars at zero or improving text contrast or simplifying the design. How could you apply that one principle immediately? And what difference realistically do you think you would make to how well someone understands the information you're trying to share.
That's a great takeaway action. Just one change can make a big difference. And of course, if you want to dive even deeper, Leroy Besler's book Visual Data Insights using SaaS ODS Graphics is the place to go.
Absolutely a really solid resource.
Well thanks for joining us for this deep dive. Hopefully you've got some new tools for your visual communication toolkit.