Financial services organisations still often face both ways at once on cloud and analytics.
They want analytics everywhere but, when it falls short of its promise, retreat.
They don’t particularly want cloud everywhere, but find themselves drawn to it like moths to the flame.
Why does analytics under-deliver? And why is cloud succeeding, despite itself?
Analytics is not well organised in many organisations
Faced with the prospect of better understanding anything and everything that moves in their particular operation, market or risk area, it is perhaps understandable that line of business executives want to get going with something small, test and learn, then
As they do so, initiatives proliferate, skills are not pooled and learning is slower than it need be. The larger the organisation, the more fragmented the picture. There’s good evidence that a dominant model of doing analytics has yet to emerge (The
Analytics Advantage – We’re just getting started, Deloitte, 2013) despite the industry skirmishing with analytics for the last twenty years or so. Centralise, and business lines complain that they aren’t getting the service they need. Federate, and costs
go up, different packages are championed and methods remain stubbornly non-standardised.
Line of business executives are not natural analytics sponsors
Line of business executives are often not that good at translating a business objective on their balanced score card (‘grow the aggregate margin by 10 basis points’) into terms on which analytics can deliver actionable insight. Instead of developing an
hypothesis which can be proved or disproved, they ask their data scientists to explore the problem and come back with ideas of how analytics could help.
So the data scientists set out on their exploration, develop ideas and, in the absence of sharper direction, steer by their best lights to produce some results which the line of business executive finds inconclusive and asks them to try again. These inititiatives
often fail to deliver substantive benefits, lose sponsor support and peter out.
Technical skills are at a premium
That good data scientists are like hens’ teeth; that there is an acute shortage; that our universities are failing to attract and produce anything like enough of them. Again good evidence supports these claims. In the UK, the candidate pipeline is well
below demand (Mind the Gap – A report on the UK’s technology skills landscape, Hired.com/skills-gap, 2016), reflected in soaring salaries for the most experienced.
Beware the Wow factor
To compound the difficulties, technological boundaries are dissolving making it possible to link any data with just about any other data. New computing languages abound whilst Business Intelligence software packages visualise datasets with jaw-dropping speed
and ease, fuelling senior executive expectation which cannot be fulfilled.
Making sense of the vortex
How to navigate to a better place? Some thoughts from McKinsey (Straight Talk about Big Data. McKinsey.com, August 2016).
It is not uncommon for big data projects to deliver only incremental, as opposed to discontinuous, benefit. And this is OK as, when you add them up, they make a major difference to the P&L.
So championing lots of small analytics efforts and measuring their impact makes good sense.
Analytics is as much about long term cultural change as rare, smart skills applied to data on clever technology. It also takes leadership from the top and sustained investment to re-orientate a firm to embed the exploitation of data.
So top-down sponsorship expressed in investment to re-orientate the business is critically important.
Once embedded as part of the culture, analytics starts to drive strategy and operations. Processes become more intelligent. Operations can progressively be automated. Customer franchises can be nurtured to yield higher returns. Risks are better understood,
monitored and neutralised before they can do much damage.
So it's a long haul.
How is this done in practice and where to start?
One authority (Analytics-Driven Organisations, Nicholas Mallison, 2015) proposes that to become an analytics-driven organisation, you need to get nine things right. Yes, nine.
The right leadership: Without strong analytical leadership analytical efforts will not gain the direction and organisation-wide support that that is required in order for analytics to be put at the heart of the organisation.
The right strategy: organisations need to see analytics as a core value creator. Therefore they need to develop an analytics strategy with a clear, shared vision and specific business targets.
The right culture: Managers and employees need to develop a belief that by making decisions based on facts they have a valuable means of validating their intuition. Where intuition is challenged by the evidence, there is an improved basis for decision-making.
The right capability and governance: an enterprise wide analytics operating model in which analytics resource allocation is governed strategically, analytical effort is governed towards business priorities and best practices are centrally collected
and shared enterprise-wide.
The right skills and competencies: Analytical skills and competencies are key sources of competitive advantage: a critical mass of such skills and competencies, well trained and educated analytics employees, a level of ‘up-skilling’ of purely technical
employees with business skills.
The right data: analytics driven organisations have a data strategy in place. They have extensive knowledge of all data sources that are around, know which data sources are both relevant and accessible per use case and have plans in place to incorporate
value- adding data sources to their existing use cases.
The right technology: leaders have an analytics technology strategy. They know how to choose the appropriate technology stack that supports their chosen business targets/use cases and data strategy. They are able to combine different technologies
to process usable outcomes and process data insights. They have a vision for emerging technologies and their integration.
The right processes and performance management: for continuous improvement analytics driven organisations have standards and policies in place to track performance of analytics usage in all phases of the analytics lifecycle. They measure and assess
accuracy and effectiveness and identify opportunities for improvement.
The right rules: analytics security and compliance policies and standards embedded in their analytics capability and processes, as security threats increase in volume and sophistication. Formalised, robust measures are required to incorporate information
security at every level.
Nobody said getting Analytics right was easy. But enough water has flowed under the bridge for us to know that a thorough, measured and strategic approach is the best way to embed analytics in the fabric of the people, processes and operations which make
the organisation perform.
So what of cloud and in particular, hybrid cloud?
And why is cloud linked so closely to data analytics?
Hybrid Cloud: asking the right question
Firms in many different parts of the financial services sector have now moved well beyond a flat rejection by IT on principle, through productive - if fragmented - use of private cloud, to experimentation and selective adoption of public cloud.
Shifts in perception have steadily followed these phases to the point where reasoned use of hybrid cloud for an ever-widening range of workloads is becoming more commonplace. Given the sheer volume of data held on customers and counterparties by Financial
Services firms, much of it Sensitive Personal Data, and the evolution of information security on the cloud, this evidence-based approach is wholly understandable.
Analytics as a Service in hybrid cloud configurations is not new. Early adoption by Marketing and Product champions demonstrated how insight could be produced and applied to improve revenue yields, cut out wasted cost. Finance, HR, Compliance then followed
suit. Then it all stated to get really interesting as business unit leaders started to grasp the possibilities.
Which brings us back to the vortex. Business and IT are not yet quite in synch with each other. What the business now wants done safely, cheaply and securely on a hybrid cloud, IT still struggles to deliver. There would otherwise be no ‘business-defined
To illustrate the lag, consider Information Week’s Hybrid Cloud survey 2014 which measured attitudes in the technical community to Cloud suitability for specific workloads, including Big Data and Analytics workloads.
Only 30% of the sample (cross-industry but with a hefty Financial Services component) claimed that it ‘could deploy on either public or private cloud, whichever makes the most sense for a particular workload’. The other 70% claimed at least to be thinking
Of those using, piloting or developing a hybrid cloud, 58% had deployed or planned to deploy more than half of workloads to private cloud. Only 6% had gone or planned to go almost all public.
On the question of Big Data (Hadoop etc.), 40% declared it best suited for private cloud, 35% equally suited for public and private cloud, and 25% best suited for public or not cloud ready.
On Business Intelligence and Analytics, 46% declared it best suited for private cloud 32% equally suited for public and private cloud, and 20% best suited for public or not cloud ready.
By way of comparison, for ERP and CRM workloads, 31% and 22% respectively declared it best suited for private cloud, 47% and 44% respectively equally suited for public and private cloud, and 22% and 32% respectively best suited for public or not cloud ready.
So some common wisdom had already started to emerge two years ago, at least amongst those engaged on the question.
Hybrid Cloud was then clearly viewed as an appropriate model by the technical community to carry many different workloads, not least Big Data and Analytics workloads. But the argument was far from over. Perceptions of the right balance between private and
public cloud were coloured by perceptions of risk.
Since then, great strides in the global supply, reach, security and richness of public cloud have been made. Microsoft and others offer boundless capacity. Some have created partner-led ecosystems of platforms and applications offered as a service. This
has led to enormous choice at every level for business users and technologists alike, which is enabling them to converge on a shared view of the art of the possible.
The challenge has shifted, as the questions of ‘what kind of cloud’ and ‘what kind of analytics’ have become so closely intertwined.
The breadth of cloud options now available mean that it has become a variable for each kind of workload. The economics of cloud are changing in parallel, accelerating the direction of travel towards greater adoption of public cloud.
Equally, data proliferation and access make the challenge for business people how best to master what is newly possible; how to test and prove value thoroughly on a bounded scope, then how scale up to reap business rewards.
The relationship between Integrators and Financial Services firms in this connection has therefore evolved into a composite of strategic business advice, domain-specific solution architecture and development, commercial engineering and service and support
delivery partner. Each of these contributes to the development of the analytics-driven organisation whose hallmarks we touched on earlier.
As these are seldom encountered in a single organisational structure in a Financial Services firm, the Integrator also plays a unique boundary-spanning role which has proved to be the decisive factor in getting cloud-enabled analytics investments to full
scale production and generate the business outcomes which led to the investment being made in the first place.
Financial Services firms have never been so outward-looking, so ready to embrace the new in order to stay at the forefront of their markets or so open to working with partners of all kinds. The next phase of hybrid cloud analytics is going to be exciting.