Predictive analytics could be the future, but we need to solve the data problem first

“Your voice breaks.” “We lost you there for a while.”

How many times have we heard or said all these things? Or how about the white wheel of endless buffering? We’ve all experienced the endless glitches, glitches, and broken app experiences that impact us more than we might care to admit.

For enterprises, the move to the cloud and reliance on SaaS apps has made the web the backbone of the business. The internet is the digital supply chain that ensures users have a great digital experience. But there’s no guarantee that your app and its many components scattered across multiple cloud environments will actually perform well.

So for today’s organizations that want to be more proactive and more automated in the way they operate and manage their environments, is delivering a predictable digital experience in an unpredictable internet environment even in the cards? I would say the answer is yes, but only if you fix the data problem first.

The data problem that is the internet

IT operations are an overwhelming place to be right now. In today’s connected world, where every business, application and device depends on a digital connection every hour of the day, driving superior digital experiences is critical. But with apps running in the cloud and accessible from many remote endpoints, the number of new blind spots has created huge challenges for anyone charged with fixing faulty user experiences. This complexity creates a network model plagued by reactive troubleshooting, and the user experience is regularly degraded.

Network professionals tell us that responding to disruptions and meeting new business needs are their top two network challenges. For these companies, the pursuit of predictive intelligence is all about the ability to move from reactive to preemptive, detecting issues before they affect the user experience. Forecasting and taking back control of what happens in the cloud has now become the core of the corporate network.

Predictive intelligence: unlocking efficiency gains and opportunities

But predictive intelligence promises real productivity gains. For organizations with a hybrid workforce, the benefits can be significant. Predictively identifying a single service impacting the outage and remediating it—for example, by switching providers and paths that carry app traffic during peak periods—can save a single agent hours of downtime or reduced performance. Multiplied over the workforce, that number quickly becomes material.

The same goes for meeting consumer demand. In the age of exponential choice, proactively preventing disruptions is key to delivering the always-on digital experience buyers need and demand. In fact, the expectations of digital experiences do rose.

Unlocking efficiency gains and opportunities to increase brand equity is the true payback of predictive intelligence.

Assessing the data-shaped challenge in predictive intelligence

Troubleshooting is a largely reactive endeavor that relies on analysis and informed decision-making to improve situations or uncover possible root causes of an active incident.

Determining what’s wrong or what went wrong meets an immediate need, but escaping that cycle of users leaving your lagging application or unavailable cloud service is no help.

That’s the promise of the predictive internet: the ability to leverage a rich dataset and visualizations to analyze historical patterns across a complex mesh of proprietary and third-party networks to predict outages or service degradation and take corrective action before the effects are felt by users.

Predictive intelligence at this level is both a data problem and a scale problem. Solving these issues is essential to making it an implementable reality.

It takes a huge amount of data to predict the onset of a degradation or performance degradation with a high degree of accuracy. While the amount of data required to train a model has been around for some time, the data was often not as clean as it needed to be. This caused flow-through effects in statistical models. Without good data, the models were simply unable to produce detailed assessments and actionable recommendations.

With modeling technology matured and backed by high-quality data collected from across a customer’s WAN network, predictive intelligence is firmly within reach.

A guiding hand

So what does predictive intelligence look like today? It starts with visibility and ends with confidence. Data-driven visibility that provides visibility into the cloud and internet environments that an organization doesn’t own, but that have become part of a corporate network and are therefore critical as a delivery mechanism for digital experiences, is critical. And just as important is supplementing that visibility with proprietary data from an analytics model that learns from past behavior and predicts future events.

Third, and perhaps most importantly, is recommending actions to take based on data and understanding of continuous performance measurement and assessment. Giving up control of the IT infrastructure is an impossible question without first building trust. Recommendations build trust. Trust that the data is correct and trust that the recommended action will produce the intended result.

Predictive intelligence should be seen as a guiding hand that helps businesses see and measure the performance of all networks that impact user experience, predict issues based on historical data, and influence decision-making.

Mohit Lad is co-founder and GM of Cisco ThousandEyes.