Shakti Mohapatra, Data & Analytics lead at Coeus Consulting, shares his advice on how to get the most from Data using cloud.
1. What are the first steps an organisation should take when embarking on using cloud to help them understand their data better – how do they work out what they want to know when suddenly there is a whole new world of ‘insights’ opened up to them?
Data is data, no matter whether it exists in the cloud or more traditional datacentres or even devices. Whilst cloud data platforms can provide horizontal as well as vertical scaling, flexibility and speed, organisations should not confuse flexibility and scale with quality of insights they are able to get from their data. In other words, a cloud platform alone cannot solve core data management issues.
We have always advised clients to start their journey from the point of having a robust data strategy and data operating model that is relevant to their scale, and more importantly think about all the insights they wish to glean from their data. We strongly believe, it is not so much about the quantity of insights an organisation enables access to; it is more about being able to access the few that make most impact on top and bottom lines and the experience of customers. Everything else is not much more than good-to-know management information.
2. Is it important to look at how data is currently structured, and consider whether structuring it differently is useful?
Definitely. This is in fact one element of the bigger data governance puzzle albeit an important one. Ensuring that data is correctly stored and structured can significantly reduce quality concerns especially for organisations that derive insights by mixing their own internal data with external data-streams. If the internal data schema and model matches external stream plumbed into data stores, it makes life much easier for data stewards and significantly reduces time to insight.
3. What about cleaning data up – what might that involve, and how can an organisation achieve it to best advantage with least time and cost? Should they trust a third party or do it themselves?
Pick up any recent survey of CDOs and you are very likely to see data quality figure very high in the list of blockers that is most likely to compromise the promise of analytics and insight-driven decision-making. That said, a lot of the frustrations stem from the fact that this issue sucks a lot of the ‘data budget’ and sometimes the outcomes are far from satisfactory. We consistently advise clients to use external help to tackle this issue and position such projects as smaller outcomes and use-case focused endeavours rather than a multi-year ‘do it all’ programme that delays benefit realisation. It’s also very important to involve ‘data-stewards’ from the business who are closest to the actual data in question because they have the best knowledge of what’s wrong with the data used to derive insights. This may sound obvious but our sampling of data-quality projects across sectors and scales provides evidence to the contrary.4. Once old data is clean, how does an organisation ensure it collects the right data, in the right way for its new system to get the most out of it?
Again, this goes back to the previous point around starting from the end and working backwards. Organisations must first decide what impact they want to drive, then look at the data they need, and develop the strategy to keeping it clean, consistent, and current and not the other way round. Once the strategy, operating and governance and data-sharing principles and mechanisms are in place, we can proceed to think about tool or platform specific needs. Most modern data platforms today, have built-in intelligence and capabilities to catalogue from a variety of workloads and engineer data pipelines with existing applications either automatically or with relatively minimal intervention.5. Is there an issue around all this with siloed data and if there is how do organisations overcome it? How, in very practical terms, does an organisation change its thinking from silos to making data available across the organisation?
Data accessibility for parts of the organisation where it can drive most benefit is often a key pinch-point, and this is invariably driven by data silos. Legacy enterprise systems that were never designed with data accessibility in mind persist in a large majority of organisations and aggravate the complexity of resolving silos. Thankfully, this is where a modern cloud data platform really helps in taking an iterative approach to centralising useful data and driving good data-driven practices. Organisations that start small and tie all data centralisation effort to key benefits and outcomes and iterate continuously over a period, get far better results than organisations that undertake it as a big-bang programme of work across the enterprise.6. AI and deep learning can have a role in getting more from data – but only after the above points have been dealt with – true or false?
True. AI and deep learning provide benefits when the data they are fed is high-quality and when organisations know what impact they want to drive from the insights they hope to create from their data.
7. Do you have any advice for an organisation that’s decided to use the cloud to get more value, meaning and insight from its data?
- Start with the impact you want to drive and work backwards to ensure your strategy, operating principles and commercial decisions are all objectively traceable to the outcome.
- Focus on establishing foundational data capabilities and data quality rigours before buying pricey tools.
- Remember that benefits from data can be realised incrementally and often do not require huge investments in a dedicated data organisation. Think big, start small and look to scale fast whilst you learn along the way.
To find out more about Coeus' Data & Analytics capabilities, click here.
A variation of this article has been published on the ITProPortal website.