How Snowflake Enables Data Unification and Large-Scale Machine Learning for Embrace Home Loans

April 7, 2021 Snowflake

Embrace Home Loans recently hosted a webinar with Snowflake to share how they partnered across IT and marketing to break down data silos, and reduced costs and delays associated with third-party data enrichment

Business data at Embrace Home Loans lived in siloes. Reporting between departments was inconsistent, and business decisions were made without a comprehensive view of data. Reporting operated like an assembly line, with data moving from one station to the next. With data moving in a serial fashion, it was difficult for the organization to see trends across the data pipeline. When faced with new regulatory requirements and a shift in consumer expectations, the company needed to adapt.

According to Dana Fortin, Embrace Home Loans Chief Revenue Officer, “We needed better communication, consistent data, and better reporting to grow and to be successful. We embarked on a concerted effort to unify our data for analytics. We wanted to use data to bring our internal teams together.”

Challenges with the legacy data platform and architecture

Embrace Home Loans’ data platform was part of a legacy topology design. 

According to Joel Kehm, Embrace Home Loans Principal Data Architect, “The biggest challenges were the inaccessibility of data and a lack of flexibility. We needed to become more flexible and agile in our operating model. We realized that our data, rather than enabling us, was actually holding us back.”

The legacy topology design created several challenges:

  • Blind handoffs in moving data caused delays in processing
  • Change management was slow and expensive
  • Error recovery took too long due to issues with data movement design
  • Many critical processes were able to be updated only once a week
  • Data was stored at a third-party provider and was difficult to access
  • Monthly costs were very high
  • Leveraging technology advancements was difficult

Documenting goals to modernize data management

Before selecting a new data platform, Kehm first documented the organization’s goals to modernize its data management. The business drivers for modernization included a need to:

  • Increase the frequency of marketing activities and campaigns
  • Adopt new technologies for advanced analytics (such as machine learning)
  • Increase the ability to adapt to a changing business environment
  • Recover from errors faster
  • Add new data and systems to increase the richness of data
  • Store broader and deeper data

Unifying fragmented data with Snowflake

Keith Portman, Head of Data Science & Analytics, knew that Embrace would benefit from modernized data management. Portman found that data resided in different applications across different platforms. This meant that Embrace was unable to:

  • Enrich customer profiles for targeting and deep segmentation
  • Optimize timing and personalization of customer engagement
  • Measure ROI across customer touchpoints and channels

Kehm, Portman, and their team selected Snowflake to bring the disparate data into a unified source. According to Portman, “Unifying data in Snowflake would make for much faster data consumption and a quicker turnaround time for analytics. A unified source means we can pull the data, run the data, develop reports, and gather insights in a timely fashion.”

Enabling large-scale machine learning

In the past, Portman ran his machine learning models on a local server that frequently ran out of memory when training very large machine learning models. The move to Snowflake solved the performance issues.

According to Portman, “Now with Snowflake and our ability to leverage the cloud, these issues are behind us. I’m able to dial up and down the types of containers I use based on my memory needs. I can train machine learning models on data sets with millions of records and thousands of attributes very quickly and efficiently.”

With Snowflake, Portman can leverage open source tools written in Python and R. The tools train his machine learning models, which Portman deploys directly through Snowflake in marketing campaigns.

Using machine learning to enable more effective sales and marketing

With business data centralized in Snowflake, Portman can train a machine learning model that generates results quickly. His models help the organization better understand the effectiveness of advertising and marketing campaigns across channels, leading to improved sales and marketing efficiency and effectiveness. 

Fortin said, “You can imagine all of our different marketing and advertising channels, with varying KPI’s. Snowflake allows us to get our data together in a single source, pull that data back, create our modeling data sets, and inform us how we do our attribution for our sales. These machine learning solutions drive our business and help us make better-informed decisions.”

Uniting the organization around data

When business data is unified in a single location, it can bring the entire organization together. According to Fortin, “Technology, Data, Sales and Marketing are now one unit. They’re no longer separate departments. Their priorities are set together. For a 37-year-old company that was built in traditional ways, data has given us a way to come together.”

View the full webinar here

The post How Snowflake Enables Data Unification and Large-Scale Machine Learning for Embrace Home Loans appeared first on Snowflake.

Previous Article
Learn Snowflake with Free Hands-On Workshops
Learn Snowflake with Free Hands-On Workshops

Snowflake University recently issued its 5,000th Hands-On Essentials badge. The Hands-On Essentials Series ...

Next Article
Who are the best SaaS providers? Those that drink their own champagne
Who are the best SaaS providers? Those that drink their own champagne

When looking for SaaS providers, factor into your assessment that those who use their own solutions act as ...

×

Subscribe to email updates from the Snowflake Blog

You're subscribed!
Error - something went wrong!