For me I use a stacked horizontal bar chart when showing the proportion of time/ money/ resources/ etc. spent in a given stage of a linear process. The bar categories are the process steps going left to right. You can quickly visualize most of your resources are going. I find it pretty intuitive and better than a pie chart in this situation.
I do not like them. I find the lower values move the upper values up and down, making it hard to get values and even harder to compare differences with the next bar item.
Great Video. I'm new to the BI space, but I came from a design background and used the same principles. I tend to convey the information I'm trying to get across and then brand it when need it. I do use the doughnut chart but not the pie chart.
Couldn't agree more with these tips. Validated my tendency to remove x and y axis titles, and even the chart titles when the legend shows what you're looking at. Another thing I'd add is this: white space is your friend. It's amazing how much better charts and tables look when you allow ample white space around them.
Like always when I hear something new like this, I need time to digest it. The chart at 10:00 was better with many of the details you removed. For example, looking at that chart of retail sales, I may want to see October sales in different years. I certainly want to visit December figures. Plus, I want to see values on your chart; I would have a problem making out 38K and 42K, which is over 10%. In your discussion of the chart's label, well, the consumers of my data are busy people with varying degrees of knowledge. The first question they often ask is, "What is the chart showing?" A big label on top tells them. With your method, they have to work it out - not good. The other issue I have is your discussion of the size. I do reports that go to many people, so they require them to be large to see. It is not what I require but what they need. I like your point about pie vs doughnut; I never thought about it, but it provides a great spot in the centre to put a label.
So glad I clicked on this video. This is the first of its kind where U learn what to consider before you start to create dashboard n not to consider yourself Pablo Picasso, this is data insights not some painting competition. Thank U Sir.
I really like these design principals - they resonate. Especially the data-to-ink ratio principals. It reminds me of this quote: "I have made this [letter] longer than usual, only because I have not had the time to make it shorter" - Blaise Pascal. I always think about this quote when I've spent hours designing, only to have a design look so much simpler than the original draft. Something simplified & cohesive can only get that way if you are clear with yourself about what you're trying to achieve or say, and to iterate to reduce the noise.
Thank you for the wonderfull explanation, as a long time analist, I apply most principles without knowing. Just one tip, when removing ink from excellent example, I typically go 1 step further. I change the chart type to a Scatter (removes a lot of ink) and add the trendline as a dotted line. Since you can see the individual months, you can also see the trend and the growth with probably less ink. Thank you and keep posting your videos!
As a 35 year veteran of teaching Excel, charts, dashboards, Power BI, Power Query, etc. I'm very impressed with your content. This was particularly useful and I'll be recommending your channel to my students. Thank you!
Thanks for the video and sharing your insights. On pie charts, I have exact same thoughts - pie (or at least the concept) has its utility but then I tend to use donut charts more to spare the eyes of the user and perhaps it looks cooler than pie chart. One thing I also do is avoid axis, especially the vertical Y-axis, and instead use data labels to directly communicate the exact value instead of user having to figure that out w.r.t horizontal line. Plus it makes things neater. One place where I don't do that is when I have too many points on a line chart where data labels can make the chart too busy and difficult to read.
Fantastic vid. I worked in the field for a long time and fought many (often losing) battles with business managers who wanted me to use flashy 3-D graphs etc in the dashboards. Having read Tufte's books and taken them to heart, it was often very frustrating to explain why there were no axes or baselines, although some of the brighter ones understood when I used the "word efficiency" model: same output from fewer inputs. FYI, Dr. Tufte's surname is pronounced TUFF-tee. Keep up the great work!
Thanks for your comment, you’re the second person to tell me about how his name is pronounced. I’d never heard it spoken before so I’m glad to be corrected 😁
I think you hit the Pie chart use right on the head. Extremely effective for helping the recipient visualize distribution of the data set. I like the “no more than 5” data sets rule as the blend of multiple smaller data sets could take away from what you are trying to convey. At that point, I like to shift and combine to only two data sets (for example our market share vs competitors) and then break out the two individual data sets separately to allow you to visually go to more detail without overloading the reader, if necessary.
If you use pie charts correctly, they are absolutely useful! I do get a minor stroke every now and then though, when I see what appears to be 314 different categories cramped into one chart. IT RUINS THE PIE! Anyway, thanks for the great video Adam!
I gained a lot from this video. It really helped highlight the need for not just substance in our dashboards, but focus on essential elements as well. Thanks!
Fantastic video. I have a problem with visualisation in general because I am a numbers guy with pretty good knowledge of statistics. I am a computer scientist though and not a salesman. What made me look for this video is that I have trouble taking people along & telling them a story & while eating humble pie I admit begrudgingly that visualisations are really important to take people along. Thanks for the video.
This is great, thank you! In essence to build a top quality dashboard takes a deep understanding of data in hand, the user(s) intention / role, prioritising lean a clean presentation
Very informative as its well grounded on literature explained in a simple way without losing substance. I enjoyed the video from start to finish. Thank you for taking the time to prepare and make it!
Great video Adam. I’ll definitely make use of Tufts 8 principles going forward. The data to ink ratio isn’t something I’d previously thought about, but your examples certainly put into perspective the distraction that unnecessary ink can cause.
Great video for beginners like me, off to watch the dashboard video ! A lot of people talk about pie charts, but the data to ink ratio was an interesting takeaway for me.
I only use binary pie charts! A yes/no or other binary split usually It's very powerful and doesn't confuse anyones Otherwise it's bar or column charts Unless your audience UNDERSTANDS the chart, don't use it!! There's more people maths-illiterate than you realise My guiding principle? Learn from advertisement posters! Very very simple messaging No more than a few items of data to soothe the furrowed brow of the viewer 🤩🌟
A useful video, thanks for sharing your insights. I heard about Edward Tufte a few years back from an Economics Professor (in the days before BI software) but his ideas are highly applicable now.
Gereat video, thanks. And as an aside, Tufte also invented sparklines... as another aside I believe his name is actually pronounced with the final 'e' so something more akin to "tuff-tee" - that's what I seem to recall from one of his presentations I attended many moons ago.
Thanks for setting me straight on how to pronounce his name. I've never actually heard anyone say it before so I was just assuming that's how it was pronounced. Great fact about sparklines as well! Data nerd gold!
@@LearnBI fully understood I'm sure I was pronouncing his name as 'tuft' as well until I went to one of his presentations. He also has a bit about information density from various media, and rails against PowerPoint-less as a not so good vehicle for information, in fact it's tantamount or at least equivalent to propaganda sources... kinda funny, maybe that's why marketing types love it - but alas maybe too true, people are supposedly only able to hold a few items in memory so after the slides clear so does the memory of them in context.
What an excellent video, I will definitely be using the tip on data to ink ratios! I like how clearly you explained each concept, I have just subscribed 😊
Brilliant video. It even changed my opinion on pie charts, but only in intuition appeal for lay people. If you're working with analytical people or trying to train them, pie charts should still be purged with fire.Even in the rare situation where there would be no other highly relevant information to add, stacked boxes are still better.
Thanks for this - good vid! I don't hate pie charts, i just think they are overused. But if you want to see the relative distribution of a small number of elements that are a part of a whole, it's hard to beat a pie chart. That said, I always start by asking myself "what question is being answered" and if there's a better option!
1:05 What was the book? I am a time traveler, so I just stared with the oldest videos :D , however, I hope that your mission is to "Saga continuous" as the way how you explain certain topics in regards to BI and Dashboarding, makes my journey much more easier. Thanks a lot and all best to you Sir ;)
Great video and advice, but I have to side with Tufte about those dumb piecharts! Add a simple donut chart, if you have space to waste and like the aesthetics, but there is always a better option
Appreciate the recommendations, particularly about the pie chart. When I was told not use pie charts, I still use them occasionally precisely because in certain cases, the distribution can quickly be evaluated. You use other charts/viz to display details on drilldowns. On the Tufte part, I actually enjoyed Stephen Few's book (Show Me the Numbers) more that Tufte's...sorry :P) And I'm with you on simplicity of design and minimal colors to draw the eye to the data story. Unfortunately, I still see too many bright color slides presented as the company standard.
Thanks, nice discussion of what is sometimes also known as the most important principle of engineering - KISS, i.e. Keep It Simple Stupid. The point is the same, elevate meaningful relevant substance over cute, superfluous style. That said, however, there IS art and craft behind making a good visual representation of anything. Dashboard or user interface design has more to it than just reductionism.
great video, at 14:30 i think that the issue there is that there is no real purpose of saying that something sums to 100%, normally what you want to compare values and it is difficult to see that comparison in a circular form.
+ so imagine you are getting votes, it doesn't add to say the people "hey, 100% of votes are here", or "if we sum all our preferences it is 100%", "if we sum all of our individual spending categories it is 100% of our spendings" because you want to know the result that data is willing to communicate, not the wholeness of it. An exception would be only with 2 values because it quickly shows the balance .
I do think that like UX and visual design elements can be a signal to senior management of information and a beautiful, clear, and ALSO functional insight is a valuable commodity to demonstrate data value. The Presidents Daily Brief is a printed, high quality, leather bound book and that communicates the value and forms part of the messaging. I absolutely approve of the dramatic simplification you outline, but I urge that managers are often real humans so how they feel about a thing can change how they perceive the data.
I also would advise that I overhauled my palette to use color blind colors, and also for many of my older managers they often noticed them squinting at graphs so moved my font and graph sizes up and that seemed to help. For pie charts, I found donut graphs as they seem cleaner.
I find 2-level pie/donught charts useful. eg: If you have 20 items that fall into 3 categories. Showing all the items of a category in various shades of one colour allows you to see the relationship of the groups together with the largest contributors to each group. Yes you can show 3 stacked bars but the pie is often more visually appealing.
@@LearnBI I agree. In my mind donuts & sunburst are just variations of pie charts with sexy names. But then I'm just a guy who sees 8 paint colours. I have no idea what persimmon & light ghost are. 🙂
Pie charts are controversial. I like using them to add some flavor to a dashboard, but I prefer not to have more than 3 categories in it (2 is perfect). If you need to display more than 3 categories, a bar chart makes more sense.
If you have many values to represent in a pie chart, what do you think about adding the top 4 and having the 5th be "other" assuming the "other" is smaller than any of the other 4 sectors?
I haven’t written book on the subject, but it seems like the author’s objective is minimization of “ink”. However, there is a certain benefit to having some ink. It makes knowing what is being plotted or graphed easier to determine without having to analyze the chart just to see what is being viewed. Contrary to that tho, I really did like it when you reduced the dates to just every year. That made it more obvious what time frame the data was representing
i find it odd that the year wasn't darker and the months lighter so that you can offer both visually but not need to print the actual months. and if not months then quarters of a year. but then again I tend to get too technical when I dont need to be.
I prefer to use pareto diagrams instead of pie charts if there are more than 5 categories. A Pareto diagrams shows in a glance the vital few categories. It’s easer to label categories rather than color coding them. Many pie charts give the feeling I’m in a DIY store trying to match the paint color. Even the example you showed with 3 categories already has similar colors between categories. Unfortunately, many people don’t understand Pareto diagrams which is a bit of down side as a good dashboard shouldn’t need explanation.
Thank you for bringing this to light, pie chart has become a staple in dashboards nowadays as its easier to show how 100% of a metric is used. Off to the dashboard video now, however, if you could answer...2 things...how often should we go for drill downs and is showing tabular data on a dashboard is fine in some case?
Interactivity should be used to help the viewer understand the data better. What options you use will depend on the dashboard's intended audience and the requirements they have. And tables are absolutely fine on a dashboard when it is the best way to present the data. Usually, it's when there are too many metrics and dimensions in the query. I like to use conditional formatting on tables as well when it can help the viewer compare different metrics.
Not to be a stickler but if you are going to remove "chart ink" make sure its not ambiguous at least. To be fair the chart was a little ambiguous to begin with...But, Management is going to look at your charts in 3 seconds and if they a basic question over the chart like, at 10:30 in the video, are you doing sales of UNITS or sales of $ over time. Which is a big difference. A single $ or axis label of units would solve this basic ambiguity that management is going to yell at you for. I don't think the principle of removing ink was to obscure the understanding of the data. Not labeling your axis correctly was my pet peeve in college. I use to have a college professor that would not label his charts, or axis and no body could figure out what the chart was until he told you what the axis were.
I'm always surprised by teacher who try to explain visual things with words and text, most of the time on video. It looks like they just read information from the internet.
However, I think there is a time and place for making engaging and exciting visualizations: Namely, when trying to interpret the meaning of the data and tell a story. You could use animations, visualized characters and places and videos for story telling. Just an idea.
This video is focused on dashboards so none of those things you mention would help the viewer better understand the data at a glance. That sort of thing can work for other kinds of as hoc reporting though.
Pie charts violate most of the recommendations you gave for bar charts. They display the data as angle, as area, and as circumference, all three more difficult for the human to comprehend than linear distance. You cannot really put an axis on a pie chart, so reading (almost) exact data is impossible. Therefore you add in the exact figures as text. Now you have the exact same data four times in a relatively large graph, inconveniently. I am with Tufte here - there is always a better way than pie charts. Also, the argument with cutting real pie is a bit bogus, as if round pie was the only thing humans ever cut into parts.
I may just be me, but the purpose of a dashboard is not to clue in someone about some interesting data or other fancy tricks. A Dashboard needs to be designed to accommodate the known needs to the viewer. The viewer should not have to divine some fact by looking at a series of lines and bars. You should know what facts the viewer needs to know and present those facts as plainly and simply as possible, even if it's just a series of Green or Red icons. The viewer really shouldn't spend more than a minute or two looking at a, "Dashboard" just like you scan the dashboard of your car only a second. Of course clicking the Red icons should take you into the Why in subsequent pages.
For me I use a stacked horizontal bar chart when showing the proportion of time/ money/ resources/ etc. spent in a given stage of a linear process. The bar categories are the process steps going left to right. You can quickly visualize most of your resources are going. I find it pretty intuitive and better than a pie chart in this situation.
I agree, that definitely does make more sense than a circular visualization 👍🏻
I do not like them. I find the lower values move the upper values up and down, making it hard to get values and even harder to compare differences with the next bar item.
Great Video. I'm new to the BI space, but I came from a design background and used the same principles. I tend to convey the information I'm trying to get across and then brand it when need it. I do use the doughnut chart but not the pie chart.
Couldn't agree more with these tips. Validated my tendency to remove x and y axis titles, and even the chart titles when the legend shows what you're looking at.
Another thing I'd add is this: white space is your friend. It's amazing how much better charts and tables look when you allow ample white space around them.
I was designing a dashboard currently and stomped across this It got me to reflect deeply on the why a dashboard is useful and stick to that. thanks.
Like always when I hear something new like this, I need time to digest it.
The chart at 10:00 was better with many of the details you removed. For example, looking at that chart of retail sales, I may want to see October sales in different years. I certainly want to visit December figures. Plus, I want to see values on your chart; I would have a problem making out 38K and 42K, which is over 10%.
In your discussion of the chart's label, well, the consumers of my data are busy people with varying degrees of knowledge. The first question they often ask is, "What is the chart showing?" A big label on top tells them. With your method, they have to work it out - not good.
The other issue I have is your discussion of the size. I do reports that go to many people, so they require them to be large to see. It is not what I require but what they need.
I like your point about pie vs doughnut; I never thought about it, but it provides a great spot in the centre to put a label.
Agree on the 10:00 details! What are the units of "sales"? Dollars? Units? Cases?
So glad I clicked on this video. This is the first of its kind where U learn what to consider before you start to create dashboard n not to consider yourself Pablo Picasso, this is data insights not some painting competition. Thank U Sir.
Glad you found it useful.
I really like these design principals - they resonate. Especially the data-to-ink ratio principals. It reminds me of this quote: "I have made this [letter] longer than usual, only because I have not had the time to make it shorter" - Blaise Pascal. I always think about this quote when I've spent hours designing, only to have a design look so much simpler than the original draft. Something simplified & cohesive can only get that way if you are clear with yourself about what you're trying to achieve or say, and to iterate to reduce the noise.
That’s a great quote!
Thanks, now ill think about it all the time too lol.
Thank you for the wonderfull explanation, as a long time analist, I apply most principles without knowing. Just one tip, when removing ink from excellent example, I typically go 1 step further. I change the chart type to a Scatter (removes a lot of ink) and add the trendline as a dotted line. Since you can see the individual months, you can also see the trend and the growth with probably less ink. Thank you and keep posting your videos!
As a 35 year veteran of teaching Excel, charts, dashboards, Power BI, Power Query, etc. I'm very impressed with your content. This was particularly useful and I'll be recommending your channel to my students. Thank you!
Thanks for your kind feedback! Much appreciated.
Thanks for the video and sharing your insights. On pie charts, I have exact same thoughts - pie (or at least the concept) has its utility but then I tend to use donut charts more to spare the eyes of the user and perhaps it looks cooler than pie chart. One thing I also do is avoid axis, especially the vertical Y-axis, and instead use data labels to directly communicate the exact value instead of user having to figure that out w.r.t horizontal line. Plus it makes things neater. One place where I don't do that is when I have too many points on a line chart where data labels can make the chart too busy and difficult to read.
Fantastic vid. I worked in the field for a long time and fought many (often losing) battles with business managers who wanted me to use flashy 3-D graphs etc in the dashboards. Having read Tufte's books and taken them to heart, it was often very frustrating to explain why there were no axes or baselines, although some of the brighter ones understood when I used the "word efficiency" model: same output from fewer inputs. FYI, Dr. Tufte's surname is pronounced TUFF-tee. Keep up the great work!
Thanks for your comment, you’re the second person to tell me about how his name is pronounced. I’d never heard it spoken before so I’m glad to be corrected 😁
I think you hit the Pie chart use right on the head. Extremely effective for helping the recipient visualize distribution of the data set. I like the “no more than 5” data sets rule as the blend of multiple smaller data sets could take away from what you are trying to convey. At that point, I like to shift and combine to only two data sets (for example our market share vs competitors) and then break out the two individual data sets separately to allow you to visually go to more detail without overloading the reader, if necessary.
If you use pie charts correctly, they are absolutely useful! I do get a minor stroke every now and then though, when I see what appears to be 314 different categories cramped into one chart. IT RUINS THE PIE! Anyway, thanks for the great video Adam!
Dashboard error 101!
Thank you for explaining in such detail for a beginner. Very helpful in sparking my interest to learn more about data visualisation.
I gained a lot from this video. It really helped highlight the need for not just substance in our dashboards, but focus on essential elements as well. Thanks!
You’re welcome
Thank you bringing back Tuft's seminal book which I think should be a required reading for Data aspiring students, or even professionals!
Fantastic video.
I have a problem with visualisation in general because I am a numbers guy with pretty good knowledge of statistics. I am a computer scientist though and not a salesman.
What made me look for this video is that I have trouble taking people along & telling them a story & while eating humble pie I admit begrudgingly that visualisations are really important to take people along.
Thanks for the video.
I was really impressed with the outcome. thank you for this insight
Fantastic, information packed, yet crisp video. Just like the way a data viz should be.
This is great, thank you!
In essence to build a top quality dashboard takes a deep understanding of data in hand, the user(s) intention / role, prioritising lean a clean presentation
Amazing... completely changed my mind about Visualization
Great, glad you liked it😁
When the report i just spent days designing is shown as horror example at the beginning .... lol Thanks for the tips it looks much better now !
Great job! 👏🏻
Which one was yours? That first one is incredible. The pizza one color use was a bit bright.
Very informative as its well grounded on literature explained in a simple way without losing substance. I enjoyed the video from start to finish. Thank you for taking the time to prepare and make it!
My pleasure
Great video Adam. I’ll definitely make use of Tufts 8 principles going forward. The data to ink ratio isn’t something I’d previously thought about, but your examples certainly put into perspective the distraction that unnecessary ink can cause.
The stacked bar chart is going to be the way to go. Pies usually create a lot of wasted space.
Excellent guide in this maximalist design era.
Really helpful
Glad it was helpful!
Great video for beginners like me, off to watch the dashboard video ! A lot of people talk about pie charts, but the data to ink ratio was an interesting takeaway for me.
Glad you enjoyed it!
Watching this for the second time this year! Really useful points. Thank you.
My pleasure!
lmao I know Im bad when the horror movie music starts and I was thinking that's a decent dashboard
Same! One of the flashes looked eerily similar to one I made, hit real hard.
Same
Lmaoooo
Same! Especially when I saw the purple design. 🤣
Same here 😂😅😅 .
This content is so necessary!
I only use binary pie charts!
A yes/no or other binary split usually
It's very powerful and doesn't confuse anyones
Otherwise it's bar or column charts
Unless your audience UNDERSTANDS the chart, don't use it!!
There's more people maths-illiterate than you realise
My guiding principle?
Learn from advertisement posters!
Very very simple messaging
No more than a few items of data to soothe the furrowed brow of the viewer 🤩🌟
Great feedback and advice, thanks for sharing 🙏🏻
If you have 5 variables, you would need 20 binary pie charts. It would not work for me.
A useful video, thanks for sharing your insights. I heard about Edward Tufte a few years back from an Economics Professor (in the days before BI software) but his ideas are highly applicable now.
Gereat video, thanks. And as an aside, Tufte also invented sparklines... as another aside I believe his name is actually pronounced with the final 'e' so something more akin to "tuff-tee" - that's what I seem to recall from one of his presentations I attended many moons ago.
Thanks for setting me straight on how to pronounce his name. I've never actually heard anyone say it before so I was just assuming that's how it was pronounced. Great fact about sparklines as well! Data nerd gold!
@@LearnBI fully understood I'm sure I was pronouncing his name as 'tuft' as well until I went to one of his presentations. He also has a bit about information density from various media, and rails against PowerPoint-less as a not so good vehicle for information, in fact it's tantamount or at least equivalent to propaganda sources... kinda funny, maybe that's why marketing types love it - but alas maybe too true, people are supposedly only able to hold a few items in memory so after the slides clear so does the memory of them in context.
I'll need to look into that, sounds interesting. Thanks for the tip.
Thanks for making this video. I feel there are a lot of takeaways that will make me consider content every time I make a new visual.
You’re welcome. Glad you got value from it. 👍🏻
Very useful video and compelling title as well. Thanks for sharing this! Great teaching skills, btw.
Thanks!
What an excellent video, I will definitely be using the tip on data to ink ratios! I like how clearly you explained each concept, I have just subscribed 😊
Thanks!
Pie charts are useful for binary comparisons like Spend to Date vs Remaining Budget, but I tend to avoid them.
Brilliant video. It even changed my opinion on pie charts, but only in intuition appeal for lay people. If you're working with analytical people or trying to train them, pie charts should still be purged with fire.Even in the rare situation where there would be no other highly relevant information to add, stacked boxes are still better.
Thanks for this - good vid!
I don't hate pie charts, i just think they are overused. But if you want to see the relative distribution of a small number of elements that are a part of a whole, it's hard to beat a pie chart. That said, I always start by asking myself "what question is being answered" and if there's a better option!
Well said!
1:05 What was the book?
I am a time traveler, so I just stared with the oldest videos :D , however, I hope that your mission is to "Saga continuous" as the way how you explain certain topics in regards to BI and Dashboarding, makes my journey much more easier.
Thanks a lot and all best to you Sir ;)
1:41 nvm, just got the answered ;)
Thanks for all content that you made ;)
Thanks a lot learned from yr tutorial exactly what I needed to make my dashboard functionally useful to users
Great, mission accomplished then!
Pie Chart when the number of values are max 5 - is a really valuable tip! thanks Adam! But the best moment stays! Maven Pizzaaaaaaaaa!!!!!!
Glad you liked the video 👍🏻
That was great! Really useful. Thank you. I've subscribed.
I completely agree with you on the importance of Donut charts. They convey a lot in a much easy to comprehend manner. 👍
Great video and advice, but I have to side with Tufte about those dumb piecharts!
Add a simple donut chart, if you have space to waste and like the aesthetics, but there is always a better option
Appreciate the recommendations, particularly about the pie chart. When I was told not use pie charts, I still use them occasionally precisely because in certain cases, the distribution can quickly be evaluated. You use other charts/viz to display details on drilldowns. On the Tufte part, I actually enjoyed Stephen Few's book (Show Me the Numbers) more that Tufte's...sorry :P) And I'm with you on simplicity of design and minimal colors to draw the eye to the data story. Unfortunately, I still see too many bright color slides presented as the company standard.
Thanks, nice discussion of what is sometimes also known as the most important principle of engineering - KISS, i.e. Keep It Simple Stupid. The point is the same, elevate meaningful relevant substance over cute, superfluous style. That said, however, there IS art and craft behind making a good visual representation of anything. Dashboard or user interface design has more to it than just reductionism.
great video, at 14:30 i think that the issue there is that there is no real purpose of saying that something sums to 100%, normally what you want to compare values and it is difficult to see that comparison in a circular form.
+ so imagine you are getting votes, it doesn't add to say the people "hey, 100% of votes are here", or "if we sum all our preferences it is 100%", "if we sum all of our individual spending categories it is 100% of our spendings" because you want to know the result that data is willing to communicate, not the wholeness of it. An exception would be only with 2 values because it quickly shows the balance .
Interesting take
I agree with you about pie charts! Also, I hadn't thought about it before that the chart represents 100% - so obvious!
I do think that like UX and visual design elements can be a signal to senior management of information and a beautiful, clear, and ALSO functional insight is a valuable commodity to demonstrate data value.
The Presidents Daily Brief is a printed, high quality, leather bound book and that communicates the value and forms part of the messaging.
I absolutely approve of the dramatic simplification you outline, but I urge that managers are often real humans so how they feel about a thing can change how they perceive the data.
I also would advise that I overhauled my palette to use color blind colors, and also for many of my older managers they often noticed them squinting at graphs so moved my font and graph sizes up and that seemed to help.
For pie charts, I found donut graphs as they seem cleaner.
I find 2-level pie/donught charts useful. eg: If you have 20 items that fall into 3 categories. Showing all the items of a category in various shades of one colour allows you to see the relationship of the groups together with the largest contributors to each group. Yes you can show 3 stacked bars but the pie is often more visually appealing.
You’d be better off using a sunburst in that case 😉
@@LearnBI I agree. In my mind donuts & sunburst are just variations of pie charts with sexy names. But then I'm just a guy who sees 8 paint colours. I have no idea what persimmon & light ghost are. 🙂
Pie charts are controversial. I like using them to add some flavor to a dashboard, but I prefer not to have more than 3 categories in it (2 is perfect). If you need to display more than 3 categories, a bar chart makes more sense.
If you have many values to represent in a pie chart, what do you think about adding the top 4 and having the 5th be "other" assuming the "other" is smaller than any of the other 4 sectors?
Yes, this is fine IMO.
Interesting. I’m always having trouble doing home data (Home Assistant) visualised in a good way. Things like temperature/humidity and such.
look at some weather apps and use the one you like looking at.
Never really thought about this, makes a lot of sense.
I haven’t written book on the subject, but it seems like the author’s objective is minimization of “ink”. However, there is a certain benefit to having some ink. It makes knowing what is being plotted or graphed easier to determine without having to analyze the chart just to see what is being viewed. Contrary to that tho, I really did like it when you reduced the dates to just every year. That made it more obvious what time frame the data was representing
What I was going to say, maybe it shouldn’t be minimization but optimization
i find it odd that the year wasn't darker and the months lighter so that you can offer both visually but not need to print the actual months. and if not months then quarters of a year.
but then again I tend to get too technical when I dont need to be.
Stephen Few has some great books on this subject as well.
For comic effect it's funnier if you do your reaction first and then show what you're reacting to.
Thanks for the advice. As you can tell, I’m not a comedian but I’ll definitely remember that for next time 👍🏻
Great video, thank you. 😃
You’re welcome
Great video and very well presented. Thank you
I prefer to use pareto diagrams instead of pie charts if there are more than 5 categories. A Pareto diagrams shows in a glance the vital few categories. It’s easer to label categories rather than color coding them. Many pie charts give the feeling I’m in a DIY store trying to match the paint color.
Even the example you showed with 3 categories already has similar colors between categories.
Unfortunately, many people don’t understand Pareto diagrams which is a bit of down side as a good dashboard shouldn’t need explanation.
Thank you for bringing this to light, pie chart has become a staple in dashboards nowadays as its easier to show how 100% of a metric is used. Off to the dashboard video now, however, if you could answer...2 things...how often should we go for drill downs and is showing tabular data on a dashboard is fine in some case?
Interactivity should be used to help the viewer understand the data better. What options you use will depend on the dashboard's intended audience and the requirements they have. And tables are absolutely fine on a dashboard when it is the best way to present the data. Usually, it's when there are too many metrics and dimensions in the query. I like to use conditional formatting on tables as well when it can help the viewer compare different metrics.
Best of Moments - the Maven Pizza dashboard!!!!
Loved it 😮❤✅️
Not to be a stickler but if you are going to remove "chart ink" make sure its not ambiguous at least. To be fair the chart was a little ambiguous to begin with...But, Management is going to look at your charts in 3 seconds and if they a basic question over the chart like, at 10:30 in the video, are you doing sales of UNITS or sales of $ over time. Which is a big difference. A single $ or axis label of units would solve this basic ambiguity that management is going to yell at you for. I don't think the principle of removing ink was to obscure the understanding of the data. Not labeling your axis correctly was my pet peeve in college. I use to have a college professor that would not label his charts, or axis and no body could figure out what the chart was until he told you what the axis were.
Great insights! Thank you.
My pleasure!
I'm always surprised by teacher who try to explain visual things with words and text, most of the time on video. It looks like they just read information from the internet.
Thank you for this!
My pleasure!
The Visual Display of Quantitative Information. Chapter 1 page 1!
Nice video!
I was taught to hate pie charts (even though I used them a lot in the early days), but appreciate your views on them.
Thanks!
This was totally excellent
Someone had to say it. Thanks for pointing this out
I love it! So helpful!
Glad it was helpful!
Great video as always
Thank you ☺️
Amazing take.
However, I think there is a time and place for making engaging and exciting visualizations: Namely, when trying to interpret the meaning of the data and tell a story. You could use animations, visualized characters and places and videos for story telling. Just an idea.
This video is focused on dashboards so none of those things you mention would help the viewer better understand the data at a glance. That sort of thing can work for other kinds of as hoc reporting though.
Pie chartwith a hole in the middle (donut chart) is bad, because the areas of circle (pie) segments become less distinguishable.
Great guidance!
Yo, for real that first dashboard was dope though
It looked great but it wasn’t a dashboard.
I do think some styling can enhance the visual experience; otherwise, we would all be just wearing white shirts.
For an ad hoc report I agree, but not so much for something as functional as a dashboard.
Very helpful!
Sometimes the clients wants the unnecessary ink.. ..or managers who think showing more is better than less, or "what if the client ask"
Very interesting video, thanks for sharing this knowledge, i think i'm gonna buy this book from E. Tufte.
Pie charts violate most of the recommendations you gave for bar charts. They display the data as angle, as area, and as circumference, all three more difficult for the human to comprehend than linear distance. You cannot really put an axis on a pie chart, so reading (almost) exact data is impossible. Therefore you add in the exact figures as text. Now you have the exact same data four times in a relatively large graph, inconveniently. I am with Tufte here - there is always a better way than pie charts. Also, the argument with cutting real pie is a bit bogus, as if round pie was the only thing humans ever cut into parts.
Thanks Adam!
My pleasure!
The most useful bits where when you demonstrated your points.
The most useful bits for you 😉
Will you make some more videos like how to choose the colour for your dashboard and how to think analytically ? Please reply ❤
amazing!!! thank you so much
Thanks for this
My pleasure!
Love the intro 😅😂🤣😍
Im also enrolled in your Looker studio course on udemy.
Thanks! You'll be one of the last because it's coming off Udemy soon and will only be available via our website.
Good job
I believe people may have different taste in Pie Chats. But I personally prefer a Cinnamon Apple Pie Chart
thank you
My pleasure
That pizza dashboard is horrible, can I have it lmao
Having worked for 2 pizza companies, I get that opening so hard
I am not a fan of pie charts and especially detest those 3D charts that Excel makes.
Long live donut charts 🍩
I rarely use pie charts, and almost despise them. A doughnut chart feels superior.
Pie charts and donut charts make me hungry.
I may just be me, but the purpose of a dashboard is not to clue in someone about some interesting data or other fancy tricks. A Dashboard needs to be designed to accommodate the known needs to the viewer. The viewer should not have to divine some fact by looking at a series of lines and bars. You should know what facts the viewer needs to know and present those facts as plainly and simply as possible, even if it's just a series of Green or Red icons. The viewer really shouldn't spend more than a minute or two looking at a, "Dashboard" just like you scan the dashboard of your car only a second. Of course clicking the Red icons should take you into the Why in subsequent pages.