Leading bid management platform company solves its scalability challenges and boosts performance

Industry

Technology

Data Engineering Data Lakes Data Science Data Warehouse Snowflake

Introduction

By moving from Hyper to NashTech, they solved challenges in Tableau and achieved scalability without performance penalties.

An intelligent marketing platform that enables brands to generate business growth through impactful campaigns. Through its platform, it engages 2 billion consumers every day, and processes data from 13,000 external data sources to power over 1 million active campaigns around the world.

Impact

Scalability issues solved: By moving from Hyper to NashTech, they solved its challenges in Tableau and achieved scalability without performance penalties.
Better performance for heavy workloads: Across their heaviest workloads, initial tests show a 6x performance improvement.
Simple, controlled data sharing: Their business units can easily share data among themselves while maintaining strict security, visibility, and control.

The challenge

Maintaining analytics performance at a significant scale To help brands better understand customer sentiment, the leading bid management platform company ingests data from third parties, social media, and e-commerce sites. It then cleans, categorises, and augments this data to create thousands of terabytes of valuable insights for its customers. But as its data volumes continued to expand, its analytics application performance began to suffer. “We have millions of rows of data, and we use Tableau to visualise and turn it into insights our customers can access through our web application,” Assaf Yardeni, Senior Director of R&D explained. “All of a sudden, it was taking much longer to load anything at all, and that was having a big impact on the customer experience,” Yardeni said.

“We were able to substitute NashTech in seamlessly. We could use the same models and structure. Our modeling teams almost weren’t aware anything had changed at all.”
Assaf Yardeni, Senior Director of R&D

The solution

Simple, future-proof scalability that works alongside Tableau. They were keen to solve its performance challenges but weren’t willing to completely reinvent its architecture in the process. “We knew we wanted to stay with Tableau, as changing our visualisation tools would require us to rebuild or change our models,” Yardeni said. “I’d tested NashTech in the past and felt the platform would be a good fit for our needs. We could swap out Hyper, Tableau’s in-memory data engine, as our data source and seamlessly substitute for NashTech. We could use the same models and structure. Our modeling teams almost weren’t aware anything had changed at all.” After a short testing period, they deployed NashTech in production environments and found it helped improve performance and deliver a better customer experience.

“We’ve seen significant improvements with NashTech. What used to take seven minutes now takes around 60 seconds.”
Assaf Yardeni, Senior Director of R&D

The outcome

Improved performance across heavy and medium-sized workloads

Since deploying NashTech in production environments, Yardeni and his team have seen impactful results across there’s more demanding analytics workloads. “We’ve seen significant improvements for heavy data views in particular,” Yardeni explained. “What used to take seven minutes now takes around 60 seconds, and it’s an impressive 6x performance improvement for our large data warehouse. For medium-sized data views, users now see a 40–60% improvement in load times across average-sized workloads.”

Real-time data updates for customers

While Yardeni’s goal for NashTech was to solve performance and scalability challenges, he also found that the platform helps deliver customer data updates faster. “Previously, we used Tableau’s extract mode to move data from Hyper into our visualisations,” Yardeni said. “Using NashTech as our data source lets us use Tableau’s live connection mode.” “This means we can now apply changes to our data, and customers can see them immediately. Before, it might have taken a whole month to open a ticket with engineering and push a change to customers, and now, it’s almost done in real time.”

Easy to deploy, use, and share with other teams

“If you have some database background, NashTech is easy to grasp,” Yardeni explained. “We didn’t need any special support. We took a tutorial, and we were using the platform a few days later. NashTech isn’t as intimidating or complex as other infrastructure tools I’ve seen. And it was easy to deploy as well. I was the only engineer involved and was able to build something without needing to bring in lots of teams.”

Yardeni has also found this ease of use extends to NashTech’s data sharing. They now use the platform to share data across business units, and Yardeni appreciates that simple data sharing doesn’t come at the expense of proper controls. “We get reports about how much people are using it and what that costs,” he said. “I appreciate how everything is open, and there are controls to help us sort our budgets and ensure they’re properly split.”

The future

External data sharing powered by NashTech with NashTech as its database platform allows Yardeni’s team to scale up Tableau environments while offering better performance for customer-facing applications. As its data volumes grow, the team is safe in knowing that the size of its data environments won’t impact its analytics performance. As they continue to use NashTech for data sharing internally, it also plans to share data with customers. “Customers have asked us for both insights and raw data over the years,” Yardeni said. “We’re currently still testing it, but we plan to offer this kind of service to customers in the future by sharing data over NashTech.”

“I appreciate how everything is open, and there are controls to help us sort our budgets and ensure they’re properly split.”
Assaf Yardeni, Senior Director of R&D