June 26, 2018
Keeping up with data technology is hard, but separating myths and marketing from reality is even harder!
Here are 5 commonly held misconceptions that we consistently encounter when talking about BI technology…
Technology is constantly evolving, and organisations are constantly turning over older technologies as they adapt and grow to the changing world around them. The odds are good that somewhere in your roadmap there is a new CRM, ERP, POS or EOM DRP that is about to drop any day now.
‘We want our new BI system to cover everything, so we really shouldn’t start that BI project until it’s finished, right!?’
Well, the truth is that trying to cover all your systems in one BI project implementation is probably not wise.
The most successful BI projects start with a single question. That question can be anything, but typically it should be centred around a key organisation metric, say for example: average dollar cost per sale. With a bit of: ‘I’d like to split that by rep and by region, please’.
Here’s the thing: the information to generate this probably only comes from a total of around 10 columns of data. So, start there and build the BI infrastructure to support those 10 columns and that single question. Treat your BI project as the ongoing journey it is and be prepared to add to it over time, injecting the data for new questions as they become relevant to you.
Just make sure that you are feeding the data for each new question into your reporting model in a consistent way and you won’t need to wait for that new system to be bedded down to get moving!
‘There’s no point reporting on dirty data, right!?’
Firstly, there is every point. Your data may not be perfect, but it is still your data and it deserves to be seen. It still has useful knowledge, so put it to work for you. Now.
Also, it’s human nature: you probably won’t clean what you don’t see.
The websites of all the popular BI tools will promote an analytical nirvana where you can just point your new dashboard product at [insert your favourite financial package here] and start building charts. Well, the truth is it’s just not that simple.
Let’s explain: You see, the tool may have the plug-in you need to connect it and it may even pull in the data effectively, but this is just the start of the process. If you want to be able to slice, dice, drill and pivot that data – and do all those self-service dashboard tricks – you will first need to construct a viable data model. Yeah, the website skipped that part ☹.
You’ll probably be able to start generating simple charts straight away, but ultimately to unlock the potential energy of your data you will need to either:
We always advise doing the data modelling in a data warehouse, rather than the baked-in data modelling tools that come with the dashboard product. With one or two notable exceptions, trying to model your organisational information in the tool itself is ultimately a nowhere road.
This leads nicely to the next myth:
They sure used to be! But what you may not know is that, these days, all the leading modern dashboarding tools use in-memory processing. ‘So what!’ We hear you ask.
Well, what this means is that Cubes are no longer required. SSAS servers and related traditional data warehousing overheads may still have their necessary place in the world, but probably not in your organisation.
And what that means is that all you need from a ‘data warehouse’ is a simple database and some SQL to load and transform the data. This significantly reduces the complexity and costs from traditional data warehousing, i.e. it can be done so much more cheaply now.
Anyone who is reading this and has known me long enough, may have heard these very exact words from my own lips. Sadly, it’s another misconception that gets passed around.
In fact, it often works the other way around – if you go ahead and just build your beast, (even if it’s a fantastic beast), people will greet it like a dose of the common cold. There are a few reasons for this, but mostly it’s because people are overwhelmed enough with technology as it is. They have been getting on OK up until now without your fancy new dashboards, so if you throw a complete BI solution into your organisation, they’ll see it as just more work!
…and the solution?
Attentive readers may have picked up on my ‘start with one question’ advice in myth 1 and adoption is another area where this approach really pays off.
The key is in asking the right questions to the right people and involving them in the process. Each iterative advancement of you BI infrastructure should touch another department, involve a different subject matter expert and embrace another perspective on your organisation. Allow people the opportunity to add to the solution as it grows and give them a sense of ownership. Then they don’t have to come: they are already there!
So now that we’ve pulled back the proverbial wizard’s curtain of befuddlement, what does this all mean for you? This is the best advice we can give: