I am working with a group of companies on Data strategy, essentially building a strong foundation and data culture, in readiness for investment in AI and ML. I wanted to share some insights on how I approach this topic, and advice I provide to Exec, Business and Technical teams when starting on this journey.
Becoming a truly data-driven organisation is not easy, no matter where you start from. We see companies adopting advanced cloud-based technologies and infrastructure, investing heavily in tooling and data science, finding they have increasingly complex architecture, unexpected rising costs, and challenges leveraging scalable benefits.
The critical component for any digital program depends on having abundant, high-quality, well-governed, and easily accessible data. This has to be underpinned by a well-communicated strategy and roadmap, where business objectives align with specific program outcomes. Building a narrative around this sort of model is an art in itself, but I always come back to the clear message being:
- Data Quality is Everything - Ensuring data accuracy is crucial, as repeated errors or uncertainty can severely impact colleagues' trust in analytics. Whilst data professionals are well aware of the need for strong data governance, the organisation may not always recognise the full extent of the consequences that governance failures can have across the organisation. After delivery and especially in the Scale-Up phases, Establishing and preserving a single source of truth is essential for all stakeholders and the absolute foundation of any digital/transformation program benefit. *(De-Duplicate Data, Embed strong environmental controls, Document data source categorisation, and labeling, Ensure security is embedded from the early architecture stages rather then retro-fitting as an afterthought)
- Keep it Simple and Share the Knowledge - Design your service architecture towards simplicity and flexibility, with 'Eyes wide-open' acknowledgment of the cost of scaling, post-project. Focus on adaptable architecture, allowing for future business needs to be met without major migrations or redesigns. While many Data Leaders aim to streamline complex architectures first, it's equally important to ensure that analytics and colleagues' use of the tools are the key measures of success. As user access to services evolves, engagement and on-demand insights will naturally increase. You must not halt momentum by losing a critical member of staff, who was a fountain of undocumented knowledge, or a supplier due to end-of-contract constraints. *(Fully document all decisions and architecture, Keep control of costs as risk against decisions, Set up strong technical governance, Validate strategic decisions against a 3 year scaling-plan, Ensure any Single point of failure is not absolute)
- Build a Narrative and Stick to Key, Simple Messages - Exec and boards understand that becoming a truly data-driven organinsation is a Journey. Ensure you have a small number of clear outputs/messages Per Stage, that will allow the organisation to see the ongoing benefit and what they get for the Point-in-time investment. Advertise data accuracy achieved through strong governance as this helps to build trust. One thing I have always gone out of my way to do is to ensure the Business is in the driving seat, delivering the benefits, NOT IT/Technical teams. *(Partner with C-Suite and their immediate managers to evangelise and assist with adoption, Taylor the message for different business functions, Make a big deal out of incremental improvements and change, Ensure you have a clear programme communication strategy/method)
- Ensure context shapes the Model - Implementing a full enterprise data architecture and governance-famework may not be a sustainable option for a small organisation. Size of the business, Delivery Skills availability, System Specialisation, BAU Change programme, Budget, and hundreds of other factors must be considered when planning a Data Transformation program. Defining a model around digital maturity helps with the narrative of what can be accomplished, at what stage, leading to budget & forecast. My Preferred method is working with High-Level 'Right to left' Programme objectives, in clear stages, with a definition of targets, using Leading Measures tatically. Maturity Level 1 Objectives = Projects 1, 2 and 3 & Timescale = Investment 'x' = Benafit 1 & 2, and so on. Digital/Data leaders must be confident to deliver a vision of the 'Art of the possible', against the resources and aspirations of an organisation. Sometimes Low-Code automation will do, rather than x Million investnet Azure Cognative services. *(Build good relationships with key cloud-service providers, Take time to understand new products coming to Market, Dont re-invent the wheel, Public sector in particular are great at sharing experience and lessons learnt, Take care and time in slecting partners to work with, Be realistic about capacity and times-scale)
