A Data Analytics Discussion

From rowing across the Atlantic to solving problems with Data. I recently caught up with James Doust to get his perspective on the Data Analytics Market and understand some of the challenges he has encountered in different organisations through his career.

What are the major trends you’re seeing in the Data Analytics space?   

 'Over the past few years, we've seen an increase in number and maturity of Automated Machine Learning tools (AutoML). Developing and managing ML models involves many steps, traditionally carried out by highly skilled and often highly paid human machine learning experts. AutoML aims to automate many of these error prone and complex tasks, freeing up analysts to concentrate on interpretation and insight and lowering the barriers to entry for those new to the field.

 According to AIMultiple, due to the relatively open attitude of publishing research results in this space, open source has produced a number of competitive AutoML tools. Similarly, there are offerings from product specialists such as DataRobot, as well as tech giants like Google's Cloud AutoML.

 With ML becoming ever more crucial weapon in a business's analytics arsenal, this is a trend to keep an eye on.'

 How do you structure Data Analytics team for success? 

 'I subscribe to Daniel Pink's three tenets of high performance: autonomy, mastery and purpose. At my most recent gig, I set up the BI team to be a self-sufficient, cross-skilled unit - able to ingest, model, secure, assess, cleanse, design, visualise and tell stories with data - all through the Azure stack. We owned the technology, the data flow, security and other governance activities, the business relationships and generation of visualisations and insight.

 Our data strategy and initiative pipeline were deliberately malleable and kept front of mind, to flex and change course in response to changing business need.

 Soft skills and attitude were absolutely crucial to the team's success - with all members encouraged to interact directly and empathically with their business customers. Co-location and building continuous improvement into our fortnightly lifecycle ensured honest, open communication - and enabled us to support each other improve. 

 The approach saw this highly motivated team quickly deliver multiple strategic and operational analytics tools used by over a third of the company every month.'

 What are the key personnel required?

 'My last BI team included an information architect, data engineer, data modeller, scrum master / business analyst who was a trained data scientist, couple of BI developers, a change manager and a trusted partner, Barhead, to dial up when our pipeline demanded it.  As the business increasingly relied on our dashboards and analysis, we each took on aspects of data governance, adding skills to our toolkit and company-wide confidence in the insight.

 The Citizen Developer is vitally important to the success of a modern BI programme and should be nurtured and actively encouraged. By taking 30 analysts from the business through our Data Literacy programme, training them to hypothesise with data, make use of PowerBI and curated data sets, we dramatically expanded the company's analytical capacity.'

 What is a good data strategy? and how do I best implement it?  

 'A good data strategy outlines your approach to managing data effectively across your organisation. It drives coherency across people, projects and teams, as well as bringing focus to master data, meta data, data quality, data security and other data governance activity. If well-implemented, it will reduce corporate risk, optimise investment in technology and people, and enable digital transformation.

 When I write a data strategy, I start with planning how to use data to deliver on the promise of the business strategy. Sometimes this involves a period of being a business detective, sleuthing my way to fully understanding the drivers and factors in the business strategy. From these use cases, we back-solve to understand the data need - is it external or internal, who owns it, and how do we get at it in a reliable way?

 Theoretically at this point we assess data governance elements such as access, quality, ethics, and stewardship. In reality I've found leaders of businesses with early data maturity need "a hook" before being willing to invest significantly in governance - and that's often when they start to rely on insights the data brings. I recommend focusing on security and access control from the off, and dialling up the rest when the time is right.

 We're then able to articulate the technology and integration needs, and the roles and required skillsets to deliver on the data strategy. The reality of course is that a strategy doesn't happen in a vacuum and you will likely already have a team partially in place. The proof of a good pudding is in the eating, and so it is with a good data strategy - delivery and communication of value early and often is of course crucial. 

 And as you deliver more, you learn more. At my last role, it worked for us to keep our data strategy deliberately short and punchy, containing only what was our 12 month focus, enabling us to refer to it often, and pivot when required, and keep it relevant. '

 What is one of the most common challenges you see in Data Analytics? 

 'An all too common challenge is dealing with legacy of an earlier Data Analytics attempt. A poorly delivered BI programme, say, has a lasting negative impact way beyond its financial cost: failing to deliver on the business case, it won't stop people manually validating untrusted data's lineage, it won't address the growing number of siloed data sources, it will crush business confidence and reduce the likelihood of future investment in data initiatives. And it's completely soul-destroying to maintain a moribund, reluctantly-used data ecosystem.

 I believe a Data Analytics programme has greatest chance of success when it is delivered using an Agile mindset and toolkit, such as seeking frequent feedback on priorities, delivering value little and often and being prepared to change direction to maximise return on investment.'

Thank you for participating in this Data Analytics discussion James.