Conclusion: As self-service data analytics and visualisation becomes mainstream – due in no small part to Microsoft’s Power BI strategy – traditional data teams within IT groups need to reconsider traditional business intelligence architectures and plan a migration to a new environment. Underpinning the new architecture must be a sharper focus on tools and practices to support data governance, which is not a strength of Microsoft’s portfolio.

Observations: The most immediate driver for organisations to adopt new business intelligence architectures is the rapid rise of self-service data analytics and visualisation. The Power BI client included with Microsoft Office is one such example of self-service analytics, and it is no coincidence that Microsoft is making basic Power BI tools ubiquitously available. Microsoft wishes to accelerate the need for a re-architecture of business intelligence services, with the goal of driving organisations to its Cloud and AI services.

Microsoft’s Power BI strategy highlights the need for data teams within the ICT groups to rethink several areas of business intelligence.

Data maturity: Enabling more users to experience data analytics increases the need for organisations to improve their data maturity. Data maturity is a measure of the sophistication of the tools and processes within an organisation for utilising data for analytics and decision making. An important consideration for data maturity is the processes for educating staff in the appropriate curation and usage of data. At higher levels of maturity, formal processes to allocate and support data stewardship are in place.

Data governance: Data governance can no longer be relegated to back-room specialists. The greater use and visibility of data enabled by self-service analytic services quickly exposes an organisation’s data quality and by extension data governance problems. There are several areas of data governance that must be examined:

• Business value story: Business stakeholders will increasingly be exposed to concepts of data quality, data lineage and tagging. Emerging tools will enable data governance to be increasingly managed by business (aka, not data specialist) users. However, without understanding the value of data governance to the business, participation by business stakeholders runs the risk of being lacklustre.
• Impact on architecture: New data/business intelligence architectures need to consider how data governance will be improved and determine new tools or services that will be needed. Microsoft – and most traditional business intelligence vendors such as Oracle and SAP – are weak with regards to federated data governance toolsets (that is data governance tools that support democratised data analytics) when compared to specialist data management vendors, such as Talend.

A strategy for AI: Organisations should recognise that the use of AI for decision support will become a key requirement of its data platforms by 2020. IBRS predicts that many organisations will embrace AI that is embedded in SaaS products. However, as machine learning algorithms become simple drag-and-drop capabilities, available within self-service solutions such as Power BI, the need to have a data platform, processes and data usage policies capable of supporting AI services becomes essential. Organisations should create an AI strategy – even if it is as a sub-strategy of its overall data analytics/business intelligence strategy.

New architecture: There are two broad approaches to developing a modern data analytics/business intelligence architecture.

The first (most common) approach is to attempt to revamp or replace the existing BI architecture. Since these architectures are generally based on data warehouse solutions with narrow sets of tightly managed data, such environments have not provided organisations with an agile environment for exploring data. Therefore, there is pressure to throw out the old technology and replace it with new vendor solutions. IBRS has seen an increase in organisations looking to decommission Oracle and SAP environments in favour of Power BI, Tableau and similar solutions.

However, such an approach will only deliver new value if new practices and principles that genuinely support self-service delivery and appropriate levels of data governance are put in place. It is just as likely that an existing BI environment can be made more agile by delivering better self-service tools and more appropriate governance over the top of existing data infrastructure.

Furthermore, established BI solutions have a deep sunk investment in reports and custom scripting. For example, the cost of redeveloping reports in a new platform can be prohibitively high.

The second approach is to create a green field data analytics environment, while leaving the incumbent BI environment in place, serving reporting needs. This approach effectively creates two architectures, one of which serves the historical reporting objectives of the business, and the other that supports self-service analytics or storytelling, predictive dashboards and potentially decision support objectives.

This approach incurs additional costs – or at least does not eliminate the licensing costs of the legacy solution, which is a key objective of Microsoft’s strategy. However, it allows for more rapid deployment of new data analytics services and defers costs associated with redeveloping reporting and data scripts.

Next Steps:

• ICT should sponsor an organisational “data readiness” maturity assessment. The aim of such an assessment is not just to identify areas where the organisation may wish to improve, but to help educate key stakeholders on the new objectives, business and operational possibilities provided by the new tools, and the need for improved data governance.
• Develop and share a data governance story to highlight the importance and value of well-governed data for the organisation.
• Regardless of which core data analytics ecosystem is selected, conduct a separate market scan for data governance tools that support the organisation’s governance models. While it is not imperative that such data governance tools are deployed immediately, the architecture should at least have the capability to see such tools implemented in the future.
• Develop an AI strategy in tandem with, or as part of, a larger data analytics/business intelligence strategy.
• Investigate the cost of recreating reports from a legacy BI system on a more modern system, but also consider which reports are still needed in an environment where users can decide the information they wish to review, either as a report or as a dashboard.
• Just because Microsoft is becoming a market leader in business intelligence does not mean implementing Power BI should be a priority, nor necessarily is it the best fit for your organisation’s data analytics requirements. When considering the role Microsoft may play in your organisation’s business intelligence, examination of the different objectives of business intelligence is required, matched with a carefully considered, potentially multi-year migration plan.