Recently I presented a talk at the SQL Server Birmingham User Group, chatting about how you interpret information and some of the best practice methods of presenting your reports. I had few questions asked at the end of the session, but got stuck on one, which was ‘Which Power BI Custom Visuals do you use and recommend?’. I was stuck for a moment and named a one or two, but I do use a few more, so here are my top picks from the Office Store.
Please note, the following list is in not any order of use, but it does seem to be split between OK Viz visuals and the rest. So out of 12 I use the most, 5 are from them, which is not surprising as they are awesome. In fact in adding the links to this post, they have released a new one, the Colour Helper. As I am super duper colour blind, it is very helpful!
This one, I have to admit, I do like the most and have used it for a number of projects and proof of concepts for customers. It allows a extra bit of visual fair to be added to a standard Power BI report, however it is report only, you can’t pin this to a dashboard, you get an error ‘Object is to big’. Normally I use it to add selector or other custom details to a report. You can use the on-line designer to add elements and then export the SVG file and load it into the custom visual. I like to use InkScape as it has more control on the things you can add. If you tag the objects, with the items in your data, it links it together and will display it. Quick note, if you do use InkScape, delete default ‘Layer 1’ so items are added to the root layer, otherwise I’ve found Synoptic Panel doesn’t like it and the visual will not work.
Slicers are both good and bad, great for filtering data, bad for performance on big data sets as they have the scan through the whole column to make sure it doesn’t miss one value. And they are also dead screen real estate. Use a visual as the slicer I always say. I’m not being figurative, do always say that. Smart Filter is a sort of free text like search, with auto predict, and you can add more than one value to it. It lives up to the name smart and reduces space taken in report layouts.
I did not know how much I needed this visual until it came out, at first I was bit sceptical, and wasn’t sure that this was even a need for it. When laying out a report I first go old school, pencil and paper, draw out a few ideas, then jump into Power BI Desktop and use this to see it the layout works on screen. It separates the layout from the data, and helps alignments of the objects see what fits and scans across the screen, and it displays the X and Y dimensions of the objects. Sweet.
Cards are great, cards with sparklines are awesome. You get that extra bit of information richness straight away, I mean if you see a card with +5%, great fine, we are up +5% this month, time for high fives and drinks all round… but if the trend has been downhill like a ski slope for the last 11 months, it adds a extra bit of visual context to the figure. You can also colour the background KPI style if needed. One thing that should be added is the ability to add a image to the background of the Card, just to add that extra bit of visual flair. If you want to do that at the moment, you have to use the Infographic Designer (see part two).
Bullet time. If you are a business intelligence pro creating reports and dashboards there is one name that you should know. Stephen Few, who wrote the book on Information Dashboard Design. He created the Bullet Chart, and this is OK Viz’s implementation of it, and it can render the visual as a single visual or multiple objects. Use this and never use a dial or gauge again.
This is the first none OK Viz one, and it’s not a visual in the sense of a chart, but what it does do is provide a textual analysis of the data and provide some insight. I’ve used this visual a few times and Business analysts and C level (CEO, COO, CFO etc) love this sort of stuff.
The free visual sends a subset of the data back to Narrative Science for analysis, so processing of the data will occur in their data centre, which I understand is in the US, so this maybe effected by GDPR, but it is only processed, not stored. The paid for version requires a little bit of set up, it needs a Linux Server (running CentOS 7 or RHEL 7 with systemd), and the application fires up a Docker container which runs the service. That container controls all the analysis, and the ports needed to pass the data to it, so you don’t have to config Apache, Ngnix etc, it handles it all, apart from the SSL certification, that still has to be handled by the web server. The best bit of it, is if you are worried by processing out of your data centre or Azure region, don’t worry its all in your network or tenancy.