In the history of IT there has never been so much data to collect, process, analyse and turn into actionable insights. For CIOs charged with enabling data and analytics strategies it can mean they need to let go of strongly held ideas about data control and embrace a ‘test and learn’ approach to new projects.
At a recent CIO roundtable discussion with Amazon Web Services and Accenture, CIOs shared their experiences of applying new data-gathering models across a range of sectors, and for a variety of reasons.
Accenture managing director for applied intelligence Fergal Murphy says businesses are applying data and analytic programs throughout the entire organisation, seeking to achieve a number of results.
“An effective data and analytics model can result in many things, such as boost revenue, improve post-sales customer care, monitor the sustainability of supply chains and even improve staff wellbeing. In short, it can benefit every part of your organisation,” he says.
AWS head of data and analytics, Asia Pacific and Japan, Cameron Price concurs with this view, noting that there is an increasing appetite among businesses across Australia to experiment with new ways to collect and analyse data.
“There are plenty of new data types and models that organisations are exploring, from full scale automation, through to targeted IoT deployments that achieve specific business-focussed outcomes,” he says.
Compass Group Australia General Manager Lea Cornelius says that the benefits of using data and analytics to enhance business outcomes for their customers are threefold.
“It’s about understanding the evolving demands of our customers so we can be part of their journey to grow and change service requirements, and identifying new opportunities to improve our internal efficiencies and optimise the services for and to our customers. It’s also about providing thought leadership to our customers through the use of data and insights,” she says.
Using IoT and analytic programs to drive better outcomes
During the roundtable discussion Pernod Ricard Winemakers IT director Simon Bennett described an Internet of Things (IoT) project designed to assist the winemaker in critical and timely decision-making. This includes from when to pick the grapes, through to the tanks they should go into (there are 3000 tanks at just one winery in the Barossa Valley).
He outlined one example where an electrical conductivity sensor was put into a wine press to find out the optimum time to run the press for. It took several iterations before the winemaker was able to use the data in a way that pinpointed the best timing.
“I think this is an excellent example of ‘test and learn’. You look at the data, you don’t find any insight, and then you keep different datasets and you make a correlation between different elements that suddenly unlock insights you’d never have seen before,” Bennett says.
“An important point around IoT and Big Data usage is that ability to use platforms to spin up in a relatively low-cost environment a proof of concept, or a pilot, that you are then able to test and learn. We’ve probably thrown away 98% of all of the things we looked at but there are nuggets of gold gurgling around at the bottom of the stream. That’s what you’re aiming to get to.”
Dealing with legacy and new data sets
While new data can unlock greater insights, it also has to be managed alongside legacy datasets and methods of ingesting, collecting and analysing data. During the roundtable discussion, many CIOs agreed on the necessity to move at “two speeds” – the slower, more deliberate approach, especially when dealing with legacy constraints, and the faster, more nimble projects involving new datasets.
“You have to do both at the same time. It is easier to ingest brand new data, test and learn, play with it and then create new insights and new dashboards and outputs. It’s much easier and quicker to do that then it is take legacy data sets and then ingest that into the new platform, perhaps re-do all of the business logic that you have built up over the last ten years,” Bennett says.
“You don’t lose the drive for data integrity and there are some datasets that I always want to be 100% accurate, and I will put time and energy into making sure that’s the case before it deploys. But there are other datasets that you can be much less controlling of, especially new datasets with the likes of IoT data.”
Bennett describes three layers of an effective data strategy – the technical architecture layer, the governance layer and the data usage layer. The role of IT in the data and analytics program is to get the architecture and the governance in place and let the business gather and apply the insights.
Skills required for today’s ‘data science’
ANZ associate director for analytics Sally Wang notes that what makes a great ‘data scientist’ is a hot topic and she identifies three key skills – analytical, technical and communication.
Analytical skills are required for statistical modelling, AI and machine learning, and data visualisation, and these people traditionally have qualifications in maths, statistics or econometrics. “They are strong on data modelling, programming (R programming, python, SQL) and good at using statistics software such as SAS, Matlab and Octave,” Wang says.
The technology skills are usually found in people with a computer science background. They understand data architecture – how data can be generated, stored, processed and cleaned, and linked together.
“In many organisations I think analytics and technology are separated into different areas or teams. But there are advantages for them to be together or having staff with both analytic and technology expertise, especially when we deal with unstructured data (such as emails and photos). It is very important for data consumers (analytical experts) to understand how data are processed,” Wang says.
Finally, it’s important not to underestimate the value of good communication skills. “It can be very challenging to transfer data and information into insights, stories and business strategies. A good data scientist with great communication skills is very valuable,” she says.
The role of the citizen developer
During the roundtable discussion, many CIOs expressed the need for ensuring business buy-in, for finding those win-win projects that produce immediate, and enduring, results.
Murphy pointed out that Accenture often works with organisations to ensure data projects are owned not by only technologists, but by specific business areas. He cited an example working with marketing executives on using data from multiple sources to continually fine-tune campaigns based on customer behaviour data.
“Working in close collaboration with one client we developed the next generation of marketing mix modelling, pulling in data that included social media activity as well as competitor analysis. As a result, they were able to reduce time-to-insight by 80%,” Murphy says.
At Pernod Ricard Winemakers, Bennett’s team have pushed the data analytics back onto the business, and in an overall company staff count of 1700, around 200 are ‘citizen developers’. These are people in the business – such as the winemaker – who are trained on how to use data dashboards so that they can find the insights themselves.
“That is a key to unlocking the value – you put it in the hands of the people that understand the data and business process. You give them the capability to analyse data and to draw insight from it in a very rapid environment. They build a dashboard, they create some insight, some different data points,” he says.
“It’s relaxing control over the things that IT is not the best at, such as understanding the data and using the data to draw insights that improve products and decision making. But retaining within IT what IT needs to do, which is the architecture, the governance around it.”
Evolving data journeys
The roundtable revealed the attendees were at various points on their data transformation journeys, but were keen to ensure momentum is maintained. Some CIOs noted that their initial months-long strategy is turning into a multi-year data and digital program.
For others, who were early in their journey, the focus is on prioritising the right data initiatives and moving at speed, through use of cloud capabilities while building out their operating model.
Meanwhile those who are further through their journeys said the focus is on deploying more advanced capabilities at scale to establish ever-increasing business benefits across the value chain.