Please critique my chart

[dmthornton]

I would like your opinion on the usefulness of this chart and how I can improve it.


http://i39.tinypic.com/2uo6zi1.png


The chart displays historical data. Let’s say it’s for monthly percentage increase in sales. It shows the following:

1) Percentage increase/decrease sales for each month

2) Periodic snapshots (e.g. quarter to date, 12 month cumulative)


Some of what I can conclude from this information is:

1) Sales are up almost 4% year to date, but only 1.5% compared to 12 months ago.

2) It’s been a good quarter so far (Oct-Nov), even better if you look back a full 3 months.

3) Sales YTD were relatively flat to worse until July, where it significantly turned around.


What I’d like to know is (any of the following):


1) Is it easy/hard to interpret?

2) Do you see any use for it?

3) How can I improve readability?

4) General opinions (don’t care if you’re brutal)


A little background:


When creating dashboards, I've keeped my charts simple so that they are easy to interpret. But I’m starting to find that in order to have all the information needed for analysis on one page, I sometimes need charts that allow you to interpret a variety of information. Using a car’s dashboard as an example, a simple speedometer tells you a lot. Not only can you tell how fast you are going, but also things like the rate of change, the equivalent speed in kph (or mph), danger level (redline), etc. I usually don’t think about all that stuff when I’m driving, but it’s there and intuitive. Sometimes a single chart on a dashboard almost needs to tell a story.


I hope to eventually put together a library of useful charts that I can use to easily string together Excel dashboards.


Thanks,


Dave

 

[SirJB7]

Hi, dmthornton!


From the humble viewpoint of a non-chart specialist, at a first glance I see that:

- I don't like the blue for the 1st and light blue for 2nd & 3rd (X)TD

- I don't like the (X)TD being of the same color of 3/12Mth

- I'm not sure about spacing of equal series' size to be the separator between TD & 3/12

- I don't like last bar of different color

- I'd have liked more having a sample of the orange and red lines aside the labels, so I don't have to look up and look down and look up

- It'd be better to choose more different colors for both lines, red & orange are close


Hope it helps you, I tried to be as critical as possible without incurring in criticism per se.


Regards!

 

[dmthornton]

SirJB7,


I really appreciate your comments. Also, you don't need to hold back. I can take it. When I start building charts and dashboards for the sake of my ego, I'm in trouble.


Your points are valid and I will take them into consideration. The chart was originally part of a "monthly" dashboard, so using a different color for mtd and the latest month (last bar) to single them out helped tie the chart to the rest of the sheet. Looking at it by itself though, I did think it looked a little off. Good to get some validation on that.


I will think about other ways to separate the groups of series as well as different color schemes. It gets challenging when you start to add a lot of elements.


The good thing is that once you build a chart, it's easy to copy and modify so that you have multiple versions.


Thanks again.

 

[SirJB7]

Hi, dmthornton!

Glad I helped you, even if it was a little bit.

Personally, I'd use two shades of blue, none dark, for the first group, and a I-still-don't-know-how different separation than that which looks a lot like another series.

I'd use too one only color (I like light green) for all second group.

And definitively I'd choose two more different colors than orange and red for the lines, adding the references aside to the labels.

Welcome back whenever needed or wanted.

Regards!

 

[NARAYANK991]

Hi ,


I don't want to critique your chart , since I don't think that was the idea of posting it in a forum like this ; instead I'd like to bounce some ideas off you.


Any presentation ( a chart is one of them ) should offer two insights :


1. Do we see any trends ?


2. Can we forecast or predict any trends ?


A chart which does not do either of these can be replaced by prose !


A business depends on quite a few pivots - tangibles such as customers , products and intangibles which cannot be projected by way of numbers , but which can probably be inferred from them.


So , if you are not a single customer / single product company , I would like to see data broken up product-wise , customer-wise , region-wise ,.... Year-wise , I think , at least in today's scenario of BI and Analytics , is passe.


Secondly , volume is one way of estimating growth ; there are so many other ways - how many of your orders / customers are repeat orders / customers ? Who are the most profitable customers , which are the most profitable products , what has been the growth across regions ? There are dozens of questions which can be asked ; if your data can answer these , I think your charts should project those answers.


Any comments ?


Narayan

 

[dmthornton]

Hi Narayan,


Thanks for the comments. They are definitely things for me to remember when designing dashboards. By the way, I do want people to critique my chart, whatever their comments may be. But please remember that this is 1 chart built for a certain scenario, not an entire set of analysis. There are definitely other charts to be built.


Anyway, I love the philosophical take. Let me share in some of that as well...


I feel the purpose of any type of analysis tool (charts, data, etc) is so you can make better decisions, plain and simple. Sometimes it trend analysis, but not always. Sometimes it's about customers and products, but not always. Sometimes time-series is passe, but not always. It depends upon (among other things) the industry, the times, and ultimately the audience. It can be many things, but ultimately the tool has to help the person using it make better decisions.


So what visualizations, techniques, or analytics will help you make better decisions? Well, the answer is "it depends". I would almost say that the only certainty is that what you present should be actionable, but even so, taking the wrong action is just as bad (or worse) than no action at all. So, again analysis must allow you to make better decisions. How will you know? To use an old phrase, "the proof of the pudding is in the eating". You'll only know when you look at a performance-based time-series chart! lol

 

[NARAYANK991]

Hi ,


What I meant by my comments was that having a dashboard present only one chart does not bring any 'energy' into a presentation ; if your dashboard is going to consist of several charts in one screen , then looking at only one chart and critiquing it is not very 'profitable' !


The screen as a whole should tell a story , so if you plan to stitch it together , fine , no problem.


Secondly , decision-making is the objective of a presentation , but as the old saying goes "one swallow does not a summer make" ! Unless there is a trend to any change ( positive or negative ) , it is difficult to take decisions ; after all every business goes through ups and downs ; in fact , that is the reason you are looking at a 3-month analysis , and a full year's analysis.


Narayan

 

[dmthornton]

I hear you, Narayan. Believe it or not, your non-critique is actually a critique (and a valuable one at that)!


And yes, I get your point. Actually, when you think about it decisions and trends are two sides of a coin. It's like cause and effect. In no particular order, a decision creates or influences a trend (or tendency) and a trend leads you to make a decision.