As you probably already know as your reading this; QlikView has an incredibly rich palette of design options. Your data can be surfaced and presented in all manner of ways with pizzazz – making it leap off the page and demand to be noticed. However, not all attention is good – if your latest chart creation is noticed only for its dizzying array of colours then it is likely that the insight in the numbers will have been lost.
So, how can this be avoided? Developing your own style and refining it takes months of using the tool. There are however textbooks available (my personal recommendation being Information Dashboard Design by Stephen Few) and dedicated blogs such as the ever excellent QVDesign Blog by Matthew Crowther.
This post however aims to give five key pointers to getting your design right.
Keep it Simple
One of the temptations when opening the QlikView box of toys is to get everything out and play. I have seen awful examples of a developer ‘trying stuff out’ and then leaving it in a production app. Classic misused features are Radar charts and animated dimensions – both are visually interesting but need to only be used in the right circumstances. Even simple options, like 3D effects and multi-coloured bars can serve to hide meaning rather than amplify it.
Stick to a Theme
This point actually covers two bases. First is ensuring that where possible there is as much consistency between objects and tabs as possible. Use of the Linked Objects feature (added in QV10) helps in this, but consistency goes beyond this. Use of QlikView theme files are also a must – ensure a theme is applied and also set as a default. This will ensure that the same fonts and colours are used throughout your app. Theme files can be used for more than just skinning a whole document – but that is a topic for a future blog post.
Use the Right Object
This sounds obvious, but it is amazing how often I come across apps where this has been got wrong. Apply the basic rules of data presentation and design here. If you don’t know what these are then get a reference book or some training. A classic mistake is the use of a line chart for non linear data. My favorite example though was use of gauges where the values did not fit in a specific range – the developer therefore calculated the upper bound as the current value plus n percent. This meant that as the selections changed the needle stayed static and just the small numbers around the edge changed value – classic! No insight gained and much space wasted on the page.
A good way of providing a varied selections of visualisations, without blowing all of the screen space is to implement Fast Changes. For those who have not discovered these an icon in the caption bar allows a toggle of the chart type. I tend to use these a lot for for showing raw values as well as a chart of those values. Although this is a great feature, be wary. It is easy to slip into poor chart type choices by just selecting multiple types. Also you need to set all properties, such as font and alignment for each fast change type you select. This leads nicely onto my final point….
It is a phrase I have purloined from Panasonic adverts, but I think the sentiment is spot on and I use it a lot when talking about QlikView design. As a developer or designer of QlikView apps you are crafting something that has to be both attractive and functional. Users will likely spend large chunks of their working lives interacting with your offering – so you owe it to them to get it right. I’ve already mentioned colours, alignment and font selection. Other common slips are leaving redudant icons on caption bars (there is never, I believe, cause for maximize icons), work in progress objects left on the screen, misplaced or oversized objects causing excess scroll bars, sloppy data modelling or loading – the list goes on. Some of these items I have covered in other blog posts, but all I will add here is to suggest you carefully consider each object you place in your app. Everything Matters.
Design Best Practice Demonstration
This video demonstrates some of the techniques described above, showing how an amateur looking chart can be shown some TLC to get it looking more presentable.
You will no doubt develop your own style, that will not necessarily look anything like my own, but the key thing is that design is given the consideration it deserves.