Yesterday while checking my website analytics reports on Google analytics site, I have noticed a new beta feature called
Intelligence. Out of curiosity I clicked on it. It took me to a an intelligence alert dashboard. Ok, lets just back up for a minute and understand what intelligence dashboard is before moving on. In the web analytics world, intelligence means is there something interesting happening on the site?. This could be information like 300% more visitors from city of New York on 3rd November or Pages on Conditional Formatting received 50% less traffic than usual from search engines on Monday. So, I clicked on the intelligence alert dashboard. And what a dashboard it is, very well thought out and designed. There are at least 10 dashboard best practices you can pick up from this and use in your day to day work. See it:
I have highlighted the important takeaways for us chart makers and story tellers. 1. Use date windows so that end-users can change the date to see different report. This can be done in various ways. For eg. in our KPI Dashboards using Excel posts, we have used scroll-bars. If you have pivot reports, just add the date to header section. Otherwise, you can also use data filters to make your charts dynamic. Band / highlight selected dates to so that users know what they are looking. This can be done using simple formulas and a combo-box control. Here is an example of conditionally banding charts in excel. Use effective colors - Google uses simple but very effective colors. [Get 73 beautiful excel chart templates and make better charts]
2. 3.
4.
Use basic charts Often we fancy ourselves to use some super-complicated-chart. Heck, Jon reviewed (more like lambasted) a 3d Square Pie Chart just yesterday. See how Google has used simple line and column charts to make the point. It is the same lesson every time folks - keep it simple. 5. Use dynamic charts - We, humans like to play. That is the single most important reason to have dynamic charts in dashboards. See how google has used the dynamic charts in this dashboard (scroll down and see the video to understand how this dashboard works). Making Dynamic Charts in Excel Comprehensive tutorials & examples. 6. Be smart with data labels: While data labels can help understand the charts, often dashboards have too many charts and thus data labels make it look cluttered. A simple solution is to use data labels conditionally. Ajay at has another good example at databison on interactive data labels. 7. Let your users customize the dashboard: This means ability to switch rows to columns, choosing how much information to see etc. 8. Highlight important information: use different font (or font size), have special area on the dashboard to display key metrics etc. 9. Show metrics by dimension: this is more common way to look at business intelligence reports. It might be a bit too much to do this kind of reporting from excel, but pivot tables can certain help you get there. 10. A good dashboard tells what is important and what is not: While we can argue that dashboards should show only the important, a good dashboard lets user customize the contents and clearly tells what is not important if it ever shows up. There is so much more beauty and design behind this google dashboard than what I can capture in a simple post. So I have recorded a small video (4 mins). Please take a look at it if you are keen to learn few more lessons on better dashboard design. 1. Find out which data you want to show in the dashboard Not all data is important. Especially when you are creating a dashboard, it is vital to provide only data and insights that are necessary to draw conclusions. A simple rule of thumb is: Your car is complex machine with thousands of parts, few micro-processors and tons of other stuff. But the driver dashboard shows only three (at most 4) data points at any time speed of car, engine heat, tachometer showing how fast your engine is rotating. Few on/off indicators that wont bother you unless you need to notice: seat belt sign, gear indicators, airbag status, batter status etc. So next time you have hundreds of data elements just use this analogy to cut down to the bare minimum and show only those. For our tutorial, the data comes from this rediff article with statistics for various test cricket players:
From the looks of it, there are just 4 things that are vital: total runs ever scored, highest score, average score per innings, Total number of centuries(and half-centuries). 2. Create one chart
First let us create a simple bar chart for the total score data. Just select the cells with total scores and click on insert chart icon and select bar chart as chart type. Now we get a default excel chart. I have used the following steps to adjust the formatting: Remove background Remove grid lines Adjust axis scaling: You may not want to adjust the axis scaling minimums. Read the follow up discussion here: Should bar charts always start at zero? Reader Poll Since the score are from 8001 (minimum) to 12027 (max) I have adjusted the axis scaling options set minimum value as 7500 and max. as 12500. Remove axis Add data labels and adjust their alignment, adjust font-scaling as well. Adjust colors and change the bar color for maximum value Adjust gap width (from 150 to 10 or something)
See this image with how the charts looked after each step:
[larger version] 3. Adjust the chart size / location so that it fits snugly inside the table with data Just select the chart and adjust its size and location until it fits inside the table. You may want to use aligning chart objects on spreadsheet trick. 4. Repeat the steps 2 & 3 for remaining charts Just copy paste the first chart you have created and change data source and scaling options. Adjust formatting if needed. Once they are all ready the dashboard should look like what you see below.
