The new analytics challenges keeping CEOs awake at night

In today’s advanced analytics landscape, more often than not CEOs are facing complex, analytics-based challenges.

Increasingly, CEO’s are asked to demonstrate that they are effectively using analytics to drive value and efficiency in their business.

To add pressure, boards are pushing CEOs to deliver results faster and often at lower costs. Today companies need to drive more value than ever before with the use of advanced analytics.

It’s not enough to have your data ecosystem in place, if not actively investing in analytics you risk getting left behind as faster players leverage not only their data ecosystem but also their analytics ecosystem better and more efficiently.

>See also: Manufacturers to prioritise data analytics

CEOs need to focus their teams to work together on capitalising on a new generation of data science-driven analytical solutions to empower their companies to drive ROI faster. Many organisations check the box with advanced discovery teams coming up with new models to improve their business but fall short when it comes time to productionise these models with IT at scale and the supporting business process changes.

It’s critical that the approach used to implement these new analytic solutions makes the best use of data science bandwidth as well as efficiently utilizing the analytic and data ecosystems.

It’s equally important that analytics become ‘operationalised’ in order to lay the foundation for advanced solutions such as artificial intelligence (AI) and machine learning. Here are some of the challenges and solutions we have witnessed.

Analytics ops: driving business outcomes

When people talk about AI, deep learning, advanced analytics, data science, business intelligence and deploying analytical models, we think about the algorithms, mathematics, and statistics behind them.

However, there is significantly less discussion surrounding how to ‘productionize’ these models to ensure that they begin to deliver ROI in production at scale. Simply put, an analytic prototype that never becomes productionised will offer little or no business value to our customers – this is where Think Big Analytics, A Teradata Company differentiates.

>See also: 5 ways to improve a data strategy

It believes there is a new way productionise analytical models. CEOs need to support new collaborative and cross functional teams to work together to automate the process of operationalising analytics in order to make the best use of their time and investments.

Teams need to focus on building new models, and not use their time on figuring out how to deploy them and keep them running. Analytics leaders of the future will use automated processes surrounding every part of the deployment process.

Seamlessly linking the analytics discovery, testing, selection, production and maintenance processes will be the capability that will separate the future leaders of industry.

At Think Big Analytics, we call this process ‘Analytics Ops’, and when deployed effectively, it impacts analytics similar to what DevOps has done for software development.

Analytics ops is about setting up your analytical and data ecosystems that make data scientists and data engineers productive, and about getting the right people, accelerators, best practices and frameworks in place to get new models productionised, deployed and optimised.

>See also: Big data projects planned in the boardroom for 2017

To remain competitive and ensure future growth in the fast-moving analytics landscape, CEOs are going to have to embrace Analytics Ops to ensure they are not left with a series of nicely crafted prototypes that never make it past the production line or continuing to invest in legacy cumbersome, time consuming and brittle analytics approaches in their business.

Speed matters: breaking silos to get analytics into production

Another major impediment to promoting analytic models into production faster to start generating business value is bringing together business and technology teams. We see IT and business departments not communicating efficiently and not having a shared understanding of the predetermined business value for new analytics.

In many instances, a company may work with analytic and data science prototypes, but often that the prototypes are built on sample set data or historical data which may not function the same in a real-life scenario.

Many encounter a lack of communication when these analytic models are pushed to IT for promoting to production: IT is frequently not familiar with the process or the code the business analytics teams have delivered, which can stand in the way of productionising the models for fear of poor performance or even worse impacting other parts of the business once promoted to production.

>See also: The 3 types of analytics set to transform customer experience

Analytics Ops is geared towards breaking these silos down internally and helping companies come up with a framework approach that guides both business and IT end to end, from insight into production and operation.

Many companies find it helpful to use a single, integrated ecosystem platform with a clear stream dedicated to everything from collecting data from the production environment and supporting data ecosystems, to building a model prototype and pushing it into production.

It’s this type of framework thinking that is the foundation of our Analytics Ops, and it’s what will break down silos and get models from prototype to production with both business and IT focused on delivering outcomes aligned with the strategy.

The value lies in simple things

A challenge CEOs face is keeping teams on track and ensuring everyone involved in building analytics solutions understands that business value is found in realizing revenue, cost, improved accuracy, eliminating risk, or optimising processes, not in creating the most academically elegant solution.

The business value of analytics must be measurable in order to feed into the continuous analytic ops mindset. More often than not, a simple analytic solution is preferable to a complex one which can be difficult to test, monitor or improve.

We try to help our customers look at the obvious areas where analytics can help, rather than the niche issues which may deliver limited business value. Delivering proven business value allows our teams to work with our customers on new areas of their strategy.

>See also: Majority of companies lack analytics capabilities – study finds

CEOs need to be asking themselves ‘can this solution be applied to deliver maximum insight, and is that insight easily applied to deliver positive business outcomes?’

Beyond this, CEOs as analytical leaders should drive their teams to engage closely with their customers to help them draw the connection between their analytical solutions and business challenges and drive their teams to focus on this customer first mindset.

CEOs need to be decisive and, when presented with the right information, make a decision to avoid ‘analysis paralysis’.

Whether analytics are being used to shape customer experience, make decisions on pricing, marketing initiatives, determine risk, predict failure before it happens, or open new markets, having the right information to make the right decision at the right time is key to taking the business forward.

The leaders in analytics ops will keep on winning and finding new ways to innovate amongst business and IT by doing the simple things really well, whereas the laggards are in danger of disappearing from the market entirely.


Sourced by Rick Farnell, senior vice president at Think Big Analytics

Avatar photo

Nick Ismail

Nick Ismail is a former editor for Information Age (from 2018 to 2022) before moving on to become Global Head of Brand Journalism at HCLTech. He has a particular interest in smart technologies, AI and...

Related Topics

Data Analytics