Do businesses need a strategy for the way they manage data and execute analytics? The answer is clear if you ask Mike Willett, an old friend and an irregular contributor to Commsrisk. Mike is now a Partner at EY in Sydney, Australia, specializing in giving advice about data and analytics to a range of clients. He is also an old soldier who fought in the revenue assurance trenches for many years, having previously worked as Director of Fraud and Revenue Assurance at Telstra. I caught up with Mike following the publication of his new EY paper (pictured), which explains how and why businesses should construct a strategy for data and analytics. Implementing AI and making money from data have become important topics for telcos, so I asked Mike what telcos should be doing, and whether his former RAFM colleagues should follow his lead by repositioning themselves as experts in this field.
Eric: Mike, your new paper is about building a data and analytics strategy that works. Could you briefly recap what it means to have a strategy that ‘works’ for the business? Is it about making more money, or is it more complicated than that?
Mike: I am fortunate to work with a number of organizations and not all have profit as their sole motive. However, even if you assume that private sector organizations are profit focussed then the data work needs to align to what the corporate strategy and line of business strategy have outlined as the direction the organization will be taking to achieve those objectives. This means we have to go deeper or else we run the risk of doing great analytics that don’t align to what the business is trying to do and so don’t add any value. For example, it may all be very exciting to do AI but if the business is crying out for basic business intelligence then you are not working with the business.
Eric: Is a large part of the strategy deciding who should be delivering the strategy? I’m thinking here about telcos like TELUS who have a data analytics team – they call it their Telemetry team – which works as an in-house supplier of data to other parts of the business, and who have to market their services to those other parts of the business in order to get the operating and capex budget to do what they do. I’m also thinking about the more general trend to recruit people who get the job title of “Data Scientist”. Do telcos need to recruit the right team and then empower them to develop a workable strategy with the available resources, or should they devise the perfect strategy and then decide how much they need to recruit or engage external suppliers to realize it in practice?
Mike: Lots of questions here. To the first point: yes, the strategy needs to outline the operating model for how data capabilities will be delivered within the business. A more centralized model like the one you have outlined with TELUS is one approach and can be viable. Similarly a more decentralized view where the business has data scientists in it, is also viable.
Each client’s strategy, culture, and ambition informs what might work best. Funding does become an issue that quickly emerges especially if you are operating as a shared service and is often one that sees considerable debate, especially as many of the capabilities are re-useable. Should the first “requestor” pay, while the laggards benefit?
In terms of what comes first, people or strategy, I would argue strategy comes first as that informs the type of people you might need. Realistically though, many organizations already have people in place in some form and consideration must be given to that. The strategy should also consider the delivery model about how capabilities (like data science) will be developed and then sustained. There are many kinds of models from full internal to full external. Without sounding repetitive, it depends on the unique circumstances.
To pick up on one point, can you get a “perfect” strategy? The answer is that is unlikely. But you can start to make choices about what you will and will not do, what you will prioritize and how you will manage risk. But to work for the business, things may need to change.
Eric: To what extent has cloud computing altered the ambition of businesses implementing a data and analytics strategy?
Mike: Cloud alone is not the only driving force behind organizations being willing to be more ambitious with their use of data and it would be hard to isolate its impact when you also have the rapid emergence of open source, new tooling, IoT, big data, competitive pressures etc that also change the ambition. I think there is broad acceptance in the clients I deal with that they can do more with data. The strategy seeks to answer the “what”.
Eric: Are telcos leaders in adopting advanced approaches to using data and analytics, or do other sectors lead the way? As somebody who used to concentrate on the telecoms sector, what would be your specific advice to telcos wanting to generate more value from the data they possess?
Mike: Generally telcos were ahead of the game as they have always had huge volumes of data to manage but others have caught up. Different sectors lead in different ways in getting value from data without there being universal truths but you will see leadership in the areas that matter most to an organization (again referencing the alignment to business). So, if you run an energy grid, there will be great case studies around network models. If you work in transport, you can will find great optimization models. If you are in government, you might find good examples around smart cities. Really the use cases are broad and very exciting.
Eric: In the paper you repeat your old motto: ‘an ounce of prevention is worth a pound of cure’. Some might say you’re biased because you were previously the Director of Fraud and Revenue Assurance at Telstra, but does this motto indicate that you believe risk managers and revenue assurance practitioners have the best instincts for implementing a data and analytics strategy, even though that strategy will extend well beyond their typical remit?
Mike: Old habits die hard. My personal view is that the advantage that fraud and RA practitioners have is that they generally get exposed to many different parts across a telco. They can understand where broken processes lead to realization of a risk (e.g. a fraud or leakage) and they know how to use data to achieve an outcome. This doesn’t mean they are uniquely positioned, of course. Generally organizations will look to areas that are perceived to very tightly align to corporate objectives – this means these strategies can often be driven by marketing or customer analytics people as well. But fraud and RA people should not be shy in advocating a role here. The challenge, as I quickly discovered, is that the way we solve fraud and RA problems is not necessarily the way we might solve all problems and we need to build a toolkit of approaches to be successful.
Eric: We have seen some professionals make the leap from revenue assurance to data analytics. Julian Hebden founded one of the first permanent revenue assurance teams when he was at T-Mobile in the UK, and now he’s the Chief Data Officer for the Australian state of Victoria, the state in which you live. More recently the BT manager who was responsible for retail revenue assurance has been appointed the Data Governance Director for the whole business. Should more telcos be looking to get more from their RA teams by encouraging them to take responsibility for data and analytics? Should the profession be thinking of ways to ease that career progression?
Mike: Maybe, it does depend on the individual. I can see why fraud and RA people gravitate to governance roles. For many, it’s familiar in terms of the language used and they are often good at negotiating across the enterprise to achieve an outcome. Equally so with data, as in the early days of RA there was not much data work being done at any scale in telcos outside the sales and marketing teams. This gave those with a familiarity of the data an opportunity to widen their exposure. An argument could be made that many revenue leakages are the result of poor data quality and/or poor process leading to a negative outcome. Good data governance seeks to address this but a little more broadly.
Eric: The new paper emphasizes the way EY works closely with clients so everyone understands the needs of various stakeholders. Is one of the biggest challenges to adopting a successful data and analytics strategy the fact that some businesses don’t have a joined-up understanding of what different stakeholders want and need?
Mike: Yes, misaligned stakeholders is a challenge but then there are almost always conflicting priorities in any organization and data is no different. A strategy though will help define the governance processes to resolve differences of opinion and this is why executive sponsorship is critical. The approach that we use is to understand the use cases for data that an organization wants, across multiple business units. We map those to the capabilities needed to deliver them and look for commonality. This informs the roadmap as we may be able to service many use cases from a single improvement initiative. Of course, this sounds easy in words but in reality is more challenging.
Eric: You really underline the important of trust, saying it is the single most important element of the data and analytics strategy. What can be done to instill trust?
Mike: There’s much written in business books on building trust. At its core, I would say it’s about listening and understanding the context people are working in, it’s about working together to solve problems, it’s about meeting your commitments, it’s about showing integrity in your dealings, and about removing your own self interest as best you can.
Eric: Staying on the topic of trust, I’ve heard RA people saying they possess their own data in their own dedicated system because they don’t trust data that comes from anyone else. Is that a flawed way of thinking?
Mike: I am not sure of any RA system that creates data – they mostly get it provided from other source systems. So I assume you mean the RA team wants “raw” data that is not subject to transformation before being provided to the RA team. I was of like mind and think this helps with RA work as you don’t introduce additional work to validate that data is right, you can focus on finding leakage. Today though, there are different technologies that challenge that approach – there are tools that can unpick the full data lineage from source to target so there are no unknown transformations. Alongside that, the need to physically move data, specific for one application (in this case RA) is generally not needed either with cheap storage allowing raw data to be dropped and accessed from multiple places.
Eric: You say the results of a good data and analytics strategy are impressive. Which business would you cite as a real-life example of how to obtain superior results from an effective strategy for data and analytics?
Mike: Of course you know I am bound by client confidentiality but to give you a throwaway line, if you look at many of the successful and sustainable businesses today, I would argue there would be a chunk of that success may be underpinned by making better decisions, informed by better data.
Eric: I have just two more questions. You’ve been a partner at EY for almost two years, and the rate of change for data and analytics has been phenomenal during that time. Will that pace of change continue over the next two years? Will the next two years be exciting, or scary?
Mike: I see no reason to expect things will slow down but that does depend on analytics having a social license to operate. We have seen a number of high profile data incidents that harm the trust in how data is being used. There is a risk of a backlash to these that could see a reduction in adoption and acceptance of technology. The future is always both exciting and scary but we live in interesting times.
You can download Mike’s paper on building a data and analytics strategy from here.