What does good look like?
And why not everyone needs to be a Google Analytics or Tableau super user
Welcome to part 3 of this mini-series on data-driven organisations. Last time we discussed the recipe and process of transforming your business by building new habits that put data at the centre of your operation. We also reflected on the importance of a coach to create the discipline around the process. But how do you know when you get to the end? How do you know you are in our ‘good’ category of data-driven business? This is what we are exploring here.
The role of leadership
We’ve already touched on the leadership aspects of this cultural shift, but suffice to say, leaders don’t need to be able to DO the tasks being done, they simply need to understand that facilitating the cross-functional teams, playing a key role in their early success and supporting them by engaging the L&D managers to ‘lay the foundations’ is a way to set up for success.
Critical roles beyond leadership
So, beyond the leadership that ‘get it’ and are supportive, what are the indicators that an organisation has been lead to a data-driven culture? What do the earliest teams that have learned these new habits, leading the way to data discovery, exhibit in their everyday progress that shows their coach that they have become autonomous?
First of all, let’s look at a minimum set of roles or functions for the team in question. To exploit data, we would suggest that you need the following roles. All organisations have the skills within them, but not very often assembled in these cross-functional teams with a distinct mandate to exploit the dataset.
- Translator – this is a business analyst or consultant. An experienced business mind that understands common patterns of behaviour by customers or co-workers sufficient to empathise with the business folks and their domain knowledge, to encode for the mathematician.
- Analyst/Data Scientist – this is the best mathematician you can find. Someone with a background in science/stats or behavioural psychology is ideal, or someone that wants to learn fast and will self-teach new statistics and modeling techniques.
- Engineer – if the data is held within legacy systems, then often it is helpful to have a code-writer or software engineer to simply access and possibly prepare, the data.
- Coach – this person will ensure the culture remains open to a respectful exchange of different perspectives, guaranteeing faster iteration of model creation and faster discovery of the data value. They will embed the winning habits from their experience.
In the ‘good’ state, these are the indicators and habits you’ll see
We’ve provided a ‘good looks like’ checklist below, by listing out both the indicators we look for and the habits which are evidently created from the indicators.
- Well organised backlog (tracks investigation, feedback, outstanding tests, etc).
- Short list of ‘Live’ data enquiries under investigation
- Detailed briefs for the ‘Live’ enquiries
- Iterative analysis logged with annotated feedback
- Evolution of a product approach to decision-making, supporting automation and scaling
- Business Analyst present to translate ‘business’ questions between data scientist and domain knowledge folks.
- Analyst improves the value of each investigation & expedites the workflow
- Weekly Reviews.
- Ongoing kanban boards for backlogs, hypotheses, and validation.
- Rapid and diverse feedback cycle.
- Business folks are familiar with detailed data briefing and valuable, rapid feedback.