Published: 17th November 2017
Author: Helen Gore
Iridium roamed the halls at Big Data LDN 2017 last week and listened to the industry buzz. As you’d expect there was insight to be gathered on every corner. A lot of what was said, and a whole lot of what wasn’t, confirmed our thoughts (and sneaking suspicions) over the last few months. Here’s our take on things….
• The whizzy shiny capabilities you see at conferences can often only be used once you have a robust data foundation with clean, governed data. You’ve got to get this right first, and it’s a long slog regardless of how many shiny tools you throw at it. In other words, don’t get too excited no matter how gleaming the end tool – there is no way around intelligent architecture and data cleaning….
• Data cleaning is STILL not being correctly anticipated and budgeted for! And, what’s worse, we are often so distracted by what insights we might be able to get out of our data that we forget to examine the data generation process itself.
• And speaking of budgets, make sure you have budgeted for and planned how to take your data project to BAU (business as usual). And then… continual development is required once a project has achieved BAU, but often this has not been thought of or budgeted for, either.
• Training users on how to use their visualisation tool (such as Tableau, Qlik or Power BI) is not enough, they also need training on how to interpret and gain insights from data. Which means you have to know how to marry your BI capabilities with knowledge of your industry, and train users within this context.
• There are no best practices yet in AI – so you need to be inspirational – there is nothing to copy from. But constraints are now often your down to your imagination rather than your budget. There are low cost, or even free, technical solutions around which are incredibly powerful.
• Data shouldn’t be seen as part of the IT department. Data is really just a facilitator – the goal has to be to improve the user’s experience. It is about being information-driven, not data driven.
• And finally, sometimes forcing your data to be stored in a structured environment can stop you from seeing the original context – beautifully demonstrated by Ursus Wehrli’s art: