Over the past two years, my cofounders and I have been meeting with telecoms all around the world, from Afghanistan to Zambia and the U.S. and finding incredible similarities across every company, regardless of the company size and region.
What we found is that every telecom shares the same business objectives:1. how do they become the leading network in their region and 2. how do they keep customers happy.
However, when telecoms attempt to adopt a big data strategy in order to accelerate towards these top objectives, they end up wasting valuable time and resources on dead end projects. This is not a unique situation - in fact Gartner predicts that through 2017, 60 percent of big data projects will fail to go beyond piloting and experimentation, and will be abandoned.
Around the world billions of people depend on their mobile carriers to stay connected. While carriers are making meaningful investments in their networks to meet demands for faster speeds and more coverage, managing and executing on big data and machine learning remains a challenge for many providers.
After hundreds of conversations with IT teams, marketers, and executives, we boiled down the top 3 symptoms indicating that a they are not ready for machine learning.
1. Telecoms who aren't good at analytics are not ready for machine learning
Companies who struggle with things like automating basic dashboards to monitor KPIs are not in a position to move to the next phase of machine learning & AI.
Harvard Business Review recently published a post about a growing interest amongst executives to leverage techniques like machine learning and AI, yet their companies are not in a position to make even basic analysis or reporting a seamless operation.
If underlying data is not standardized with automated processes of getting data into a unified state for analysis, then a company's ambitions to be more data driven or launch into machine learning projects will be very unlikely to succeed.
“We just realized that last month’s report had major inconsistencies, but nobody knows why. We’re afraid to tell the CEO” - Marketing manager at a major telecom
2. More than 70% of time is spent on pulling, prepping, and cleaning data
In many organizations, end users spend more time negotiating with IT and BI and preparing data in spreadsheets instead of doing what they were hired for, which is to act on key insights in order to meet critical business objectives.
Companies facing these data related issues are lacking automation necessary to transform data into a standardized schema that everyone in the organization can access and understand.
We found that many organizations are stuck in a perpetual cycle of passing data requests back and forth between IT and business teams, and dedicating more resources to clean up data issues. By the time the analysis is actually created it’s already time to move onto the next objective.
3. Telecoms have no single source of truth
Telecoms everywhere are attracting more users if the amount of data they’re capturing about their customers and networks continue to grow without a proper plan around how to store, manage, and utilize that data, then every new project is at risk of falling apart. Data is siloed, queries take hours or days to complete, and nobody in the company has a complete view into what’s actually happening in the business.
In some cases we found that business teams like marketing had gotten so desperate for access to data critical for their roles that they taught themselves how to setup their own databases and taught themselves SQL.
In other cases, we found that IT teams are lucky if they can take a few days off for vacation without putting the entire organization at risk. The big secret amongst IT and BI teams is that projects around cleaning up data issues is never ending.
While it’s amazing to see the initiative and problem-solving mindset teams take to create their own workarounds, it’s an alarming indicator that foundational breakdowns in a company’s infrastructure is the real problem to fix.
If so many telecoms are facing these challenges, this makes machine learning & AI a distant reality. So, what is the right way forward?
At Adazza, our core belief is that any telecom in the world should be able to access the same abilities as a Silicon Valley tech company to know exactly what’s going on in their business with reliable analytics, and what will happen in the future. Continuing to manually manage the data pipeline problem is not only unrealistic, it is an impediment in a company’s ability to do anything with machine learning and AI.