Three Important Trends in Business Insight for 2020 and Beyond
How we use analytics for business insight in 2020 -- and beyond -- will be transformative, making things that are impossible today possible tomorrow.
- By Anupam Singh
- December 16, 2019
As we enter 2020, customers are telling us about three important focus areas in data and business analytics. First, I see many organizations facing challenges in managing their cloud budgets, with many already dealing with the fallout of runaway costs. Second, I see a much greater importance in protecting customer data, as trust becomes a differentiating factor between retaining or losing a customer. Third, as predictive analytics becomes a standard of excellence, we need to operationalize machine learning and artificial intelligence, making them easily accessible to all data consumers, not just data scientists.
2020 Trend #1: Smart management of cloud spending
Cloud budget management will become a major focus in 2020 and beyond, receiving direct attention from both CIO and line of business (LoB) managers. As the LoBs start adopting technologies on their own, cloud budgets are obscured as operational expenses for each LoB. However, when the final bill is aggregated, CIOs get sticker shock. CFOs get involved, and there is a direct reaction to reduce cloud expenditure. This may be the wrong reaction.
As more organizations become data focused, people will naturally do duplicative work. Instead of pure cuts, we need to become smarter about keeping track of who uses what data and when, and proactively reduce and remove duplicates from the process, aggregating across functions and business units. This reduces what we pay for services in the cloud without reducing what we're doing with it. In fact, this will help us expand how much we can do while managing how much is spent on cloud technology.
A major financial services company we work with had an on-premises environment that was operating at around 30 percent utilization on weekdays, down to near zero on weekends, yet had frequent bursts of activity that took them to a full 100 percent. Moving to the cloud allowed them to stop wasting idle capacity. However, when those seasonal and rogue queries hit, the ones that would take them up to 100 percent, they ended up facing cloud consumption bills that went beyond their budget. The problem was not the cost itself. The problem was that they had no intelligent cloud workload management that would help them figure out a better utilization strategy.
You can’t fix what you can’t see. Adding appropriate tools and processes provided visibility into their workloads and enabled them to continue to run all the same queries, consolidated and optimized, and keep everything under budget, all while improving performance. Turning off the tap was not the answer. Making the service work best given that it was in the cloud made all the difference.
Bottom line: just because you can spin up resources in the cloud, do not throw away years of disciplined process. Operational excellence is still important even if you do not have rack-and-stack hardware. Optimize your cloud use to maximize the value of your cloud spending.
2020 Trend #2: Building brand trust through data privacy and security
When it comes to protecting your customer's data, we see a huge shift in awareness from the public. Recent events have placed a huge light on the value of data security and privacy protection, and the public has grown to be much more sensitive to how the companies they interact with share and use data. This will create pressure on CIOs and CSOs to protect data and make sure they keep track of who uses what data when.
This is actually a good thing. As consumers become more aware of the value of their data, CIOs and CSOs will benefit by creating a strong trust relationship with them. Trust will become a currency for banks, healthcare, telecoms, retailers, and others, as consumers make purchasing decisions in favor of brands they trust as stewards of their data. For example, if I need financial services, I will make it a mandatory buying criteria to go to a bank that has established trust with me personally in their data privacy and security protections.
This is also important in healthcare. Many large hospital chains know using data analysis to reveal patterns is the best way to predict and prevent infections, one of the largest risks of being an admitted patient. They are using data from various sensors and devices in the hospital, even Fitbit-type devices, to predict where the next infection will occur and how to prevent it. I know of one large hospital chain that is seeing thousands of cases being detected, allowing them to block the spread of infections immediately. However, this data is also sensitive. Data stewards need to make sure that only authorized people have access to data, and the data is masked where necessary.
This isn’t limited to just hospitals and banks. Every brand in the world -- sporting goods companies, telecoms, automotive companies, shoe companies, and anyone that collects data, manages data, and supplies services based on that data to their customers -- will have a major trust deficit if they do not have a clear privacy policy. It is no longer sustainable for companies to simply wait for the breach, as this will cause damaging loss of trust. All companies will suffer major brand damage if they do not become proactive and create data privacy protection and security as a top value for their customers. Enterprises need to adopt a governance platform to build a privacy framework, with auditing, lineage, authorization, and privacy as more than just empty words.
Bottom line: ensure clear data authorization and governance is in place. You must add privacy and security values to your brand and be prepared to back them up. If you do, your brand will increase in value in customers’ eyes.
2020 Trend #3: Delivering real business insight with operationalized machine learning
Business insight is not just counting things to keep score -- it is finding meaning in the myriad of patterns and correlations that emerge over time. We can automate the discovery of these patterns and correlations with machine learning, making the road from data collection to true business insight significantly shorter. For this to happen, machine learning (ML) needs to become operational and start seeing the same needs for operational excellence. ML operations will become a whole new set of CIO-focused, mission-critical technologies.
ML is ingrained in everyday things. Today, GMail practically writes all my messages for me. ML is embedded in banking to make digital assistant interactions personalized. Banks are also integrating digital assistants with brick and mortar for a much more personalized and rich experience. ML is embedded in the telecommunications market. In Asia, most subscribers are not on contract, freeing consumers to change their subscription relationships every 3-6 days. Asian telecom companies use ML to develop a high-touch relationship with their subscribers and predict the best way to keep them on their service in real time. There are hundreds of examples like this in every industry today.
When ML is embedded in this way, treating it with the same discipline and process becomes even more important. Imagine you’re using ML for credit approvals and suddenly there is a change in the ML model’s behavior. For the consumer, the ML model starts giving a radically different credit score for an individual, leading us to incorrectly deny services to a worthy customer. When the consumer calls a helpline number asking about the change in her credit score, it's important for the CIO to have put in place operational tools on when and how the model was last updated. .
Bottom line: fall in love with ML and think of it and track it like any other mission-critical production service. Support it as such. Do not let it run outside the established norms. Only then can you operationalize ML across your entire organization.
A Final Word
I truly believe that the way we use analytics for business insight in 2020 and beyond will be transformative, making things that are impossible today possible tomorrow. However, IT teams will have to build new muscle to say yes to the needs of the business. To get it right, we must be smart on how we optimize our cloud computing investments, applying the same discipline and focus we have to our historical rack-and-stack infrastructure. We must also enrich our brand value with real customer privacy and security, treat their data as valuable as we treat their business. Finally, we must incorporate machine learning as mission critical, complete with the checks and balances we provide any essential operational service in our enterprise. If we do these things, we will be ready to take on this bright and amazing future.