Search DAS42.com

Get Started
Background

Zayo provides the fiber connecting companies and data centers, plus the hardware and software infrastructure at both ends of that fiber.

The company has grown through 50 acquisitions since 2007, and now operates in 21 countries, connecting more than 1,900 data centers worldwide and managing over 224,000 route miles of fiber across North America alone.

Between 2021 and 2023, Zayo acquired three managed service providers (MSPs) to expand its offering and differentiate in the market. Each Edge MSP business unit came with its own processes, service offerings, and data infrastructure.

DAS42 and Snowflake are instrumental in customer churn prevention, revenue forecasting, and financial modeling.

By partnering with DAS42, [everyone from] the executive team to our engineers all gained measurable efficiency by leveraging data insights for impactful business decisions. Our collective team saves hours weekly by implementing DAS42’s holistic data governance strategy for our first-, second-, and third-party data.

— Frank Cittadino, former CTO, Zayo
The Challenge

Fragmented data across acquired businesses.


Zayo’s data science team had already built a predictive churn model for the company’s traditional fiber business. That model didn’t extend to the newly acquired MSP business unit.

Each of the MSP entities ran on different systems and defined retention and churn differently. Without a consolidated view, the Edge MSP business unit couldn’t see churn in the aggregate and couldn’t act on it.

Building one model meant coordinating data infrastructure, domain owners, and business processes across all three organizations at once.

The Solution

Starting with the people who know the business.


DAS42 treats every problem as a business problem first and a technical problem second.

The team ran joint discovery sessions with sales, customer success, technical success, and other subject matter experts from all three MSPs to identify the KPIs and indicators most likely to predict churn in managed services.

That work produced a feature list built for both predictive power and organizational buy-in.


One pipeline, one view

Each MSP team curated its own data for the agreed features and uploaded it to Azure Blob Storage. DAS42 built data pipelines, orchestrated with Airflow, to combine the datasets into a single view in Snowflake. That consolidated view went straight to Zayo’s data science team, who used Dataiku to train and compare multiple models against it.

Principal component analysis identified the features with the strongest predictive power, and DAS42 refined the pipelines from there.


Built to scale with the next acquisition

The pipeline now ingests data from each MSP automatically, merges it, and feeds the consolidated dataset to the Dataiku model for monthly churn predictions. The framework was built to absorb new acquisitions without a rebuild, matching the pace at which Zayo grows.

The Results

Predictive churn insights enable proactive customer retention.


Zayo can now generate predictive churn insights for its MSP organization instead of managing each business unit’s retention blind. The model reached over 90% accuracy and precision, giving Zayo:

  • Deep predictive insight into how projected churn affects top-line revenue and bottom-line profitability
  • Early visibility into at-risk accounts, so customer success and sales can act before a customer leaves

With a model designed specifically for the MSP business, Zayo has the accuracy to increase customer lifetime value, decrease customer acquisition cost, maintain revenue streams, and improve customer satisfaction and net promoter scores through proactive customer care.