At Data Monsters we talk about Dashboards a lot. You probably hear them mentioned in business (and technical) conversations at work. Often it's slipped into a sentence "We need that in a dashboard" or "Can we set up a weekly dashboard for the sales results?". Many people use the term Dashboard to mean any display of data - even if it's 50 rows in a table!
We like this definition:
A graphical summary of various pieces of important information, typically used to give an overview of a business.
It's also really important to remember where the word Dashboard comes from. It's origins are in your car instrument cluster or in a cockpit of the plane. The person driving - or piloting - should be able to see at a glance how they're doing. If you can't, what you have isn't a dashboard, it's a visualisation of data. The two are not the same.
With that in mind, here are 5 things a dashboard should be:
1) A dashboard should be visually appealing
This doesn't mean many colours, fonts and whizzy things. It means well-designed, clear, and something you want to look at. It should make use of white space, and use good design principles - similar metrics should be clustered together.
2) A dashboard should be tailored for the audience
I'm a marketer. My KPIs have typically been cost per lead, cost per acquisition, revenue per member, average lifetime value, Net Promoter Score and conversion percentage. These are great for me, but would be meaningless for a HR director who needs to understand time to hire, time to fill, turnover year to date and staff engagement. It should go without saying, but make sure dashboards meet the needs of your customers, or they won't get used.
3) A dashboard should be drill-down
We've seen it a lot. "Dashboards" that are actually emailed snapshots of a table. You can't drill down or understand the underlying data. The end user may aswell print a sheet from excel and be done with it. This is an old school way of reporting and it belongs firmly in the past. If your end users can't drill into their data from the dashboard, to ask and answer basic questions, you're doing it wrong.
4) A dashboard should combine multiple data sources
This one is debatable, but we think it's essential. Most decent software programmes these days will have a decent reporting front-end built in. If your dashboard isn't adding value over and above what the end user can get from their individual programmes, it's no good. You're wasting your time.
5) A dashboard should be secure
Data is one of the most valuable assets a company has, and with changes to legislation in the form of GDPR, it's more important than ever that controls around who can see what are tightly managed. For that reason, many dashboards aggregate data - rather than the end user being able to drill down into personal data of individuals. We think this is sensible. Tools like Microsoft Power BI have fully granular access control. It's not the sexiest thing to think about, but it's one of the most crucial.
And 5 things it shouldn't be:
1) Something you print out and leave on a desk for your boss. Really, save the trees!
2) Something nobody uses. Most applications have built-in usage stats. We think these are worth looking at.
3) Something finished. Dashboards should evolve as businesses do. You'll never be "done" while people continue asking questions.
4) Something confusing. We know visuals look cool, but you can go over the top. Remember, follow good design principles.
5) Something you can't access remotely. If you have to be hooked up to an office network to get it, it's not good enough. Many people travel, and often they're the ones who need data quickly. Their needs should be considered when you're designing and building your dashboards.
Need help with your dashboards?
If you're interested in Data Monsters conducting a review of your current data visualisations and dashboards, please get in touch!
You May Also Like:
Jo and I decided to set up Data Monsters after a visit to SQLBits in April this year. It's one of th...
A product owner is crucial for the success of the product - it must be a real commercial success - a...
I have worked on data warehouse projects in many different environments for over a decade. None have...