I’m a bit of a data nerd, and especially a data visualisation nerd. I love graphs and charts, but not for their own sake. What I love about them is the way that a single picture can show so many different things, and often provide new business insights into data. It just makes everything clearer.
Well, it makes everything clearer to me. But as a visualisation nerd, I also understand that not everyone can read graphs and charts easily. Data scientists can get a bit over-excited with the sheer beauty of their data visualisations, and start making them even more complex. They forget that the proof of the pudding is in the eating. In other words, a graph is only as good as the messages that people take from it. But how do we ensure that the interpretations are ‘correct’, and also that the right messages are shown in the first place?
Co-creation processes and agility
Historically, in my view, the best data visualisations and dashboards are those that have been co-created by analysts and business users. A process of agile co-creation means that the visualisation tends to answer the right questions. That is the ones that business users want answered, and therefore meet the dashboard’s key purpose.
The rise of self-service analytics, however, has taken this co-creation process to a new level. Citizen data scientists, or business users who are able to handle self-service analytics packages and create their own insights, are on the increase. And in citizen data scientists, the co-creation process comes together in a single individual: Those who understand both business and analytics. They know what question they want answered, and they know how to answer it and display the answer. It is, effectively, the ultimate in agile.
Well, kind of. It could be. But too often, it’s not, because citizen data scientists are not fully equipped, and neither is the organisation.
The requirements for agility
Citizen data scientists need the right tools to be truly agile, and the organisation also needs the right mindset. In practice, this boils down to understanding the importance of time: That agility is all about making things happen faster, while still ensuring that decisions are driven by accurate information.
Three main elements
First, self-service analytics needs to be driven by data that has been prepared and validated. The process of data governance is not easy. Quality assuring data needs both IT and business users involved, because business users are more likely to spot problems. But IT teams have to take responsibility for data quality overall. But it must also be done in a timely way. Data is at its most useful when it is fresh. Every day that is taken to clean it and assure it makes it less useful. There is a balance to be found between perfect and usable. Businesses need to work out what point is ‘good enough’ for their purposes.
Second, self-service tools need to provide what users want and what they are capable of using. They must be flexible, but also easy-to-use, so that everyone is able to ‘have a go’. There is a danger that if the tools provided are not up to the job, then users will look elsewhere. Perhaps there is even more risk that users will simply go back to relying on ‘gut instinct’. In practice, organisations may well find that the best option is to look to the cloud for solutions. Cloud-based analytics tend to have more flexibility to add more options if needed. This may mean overcoming security issues, and managing performance, but a number of organisations are now realising that this may be the way forward for them.
Finally, the organisation itself needs to be agile enough to use the results of self-service analysis quickly. A decision driven by real-time data is likely to be meaningless if not implemented for several months. Just as data decays over time, so do the decisions made as a result. This is likely to be a matter of culture and mindset, rather than use of technology. Readiness to accept new ideas and try things out is likely to be crucial.
Data visualisation is a wonderful thing. But to support decision-making it must show the right things, be easy to understand, and be accepted by those involved. Self-service analytics will not be enough on its own, but it is a vital step towards a more agile approach to decision-making.