Epsilon

Epsilon Unlocks Savings with Duckbill’s Expertise

Epsilon is a leader in global adtech and martech company working with the world’s top enterprises. To help streamline their AWS architecture and reduce their AWS costs, Epsilon turned to Duckbill. The savings Duckbill found exceeded Epsilon’s estimates by more than six times, and the initial findings Duckbill found paid for the consultation costs within weeks. Now Epsilon plans to work with Duckbill as they move further into serverless architecture. 

Overview

• 20MM customer database 
• 100M emails sent annually 
• 15T in transactions 
• Saved 6x initial Duckbill projection 

Managing Security, Scale, and Spend on AWS

Epsilon helps enterprises understand their customers on a granular level so they can message those customers more effectively. Their extensive platform lets companies centralize their data, automate and track marketing campaigns, and develop deep analytical insights on how those campaigns are running so companies can tailor them to their business goals. 

Scale, security, and performance are paramount to Epsilon. With Fortune 500 companies on their client roster, they have to be able to rely on the cloud services that power their platform. 

But, scale and performance can come at a high cost. Epsilon’s wide range of SaaS products, managed services, and programs that power their customer’s marketing efforts all rely on AWS. To ensure their customers have a consistent experience using their platform, Epsilon invests heavily in AWS performance. 

Epsilon wanted to crunch the numbers, review their architecture, and make sure they were getting the most out of what they were spending on AWS. That’s when they turned to Duckbill. 

The Power of the Outlier

Jason LeBaron was familiar with Duckbill before Epsilon started working with Duckbill. As VP of Solutions Architecture at Epsilon, he had seen Duckbill content floating around his corner of the internet. 

“The work Duckbill produced was meaningful and relevant to the community on the security, operational stability, and cost front,” said LeBaron. 

But, being interested in AWS cost optimization content, and being interested in working with the company behind that content are two different things. Epsilon had previously worked with a third party AWS expert that promised to cut their costs.

“The other providers asked for high time commitments, or asked to embed additional agents and data collection tools to collect telemetry and the reports we were seeing were all the same. They all said optimize CloudTrail, NAT Gateways — all these common bits of advice. We already knew this. Where was the value in telling us stuff we already know?”

“It left a bad taste in our mouth, so we were hesitant to do it again. But, Duckbill’s content and the reviews we saw from the community made Duckbill an outlier,” said LeBaron. 

Part of what drew LeBaron to Duckbill was the fact that Epsilon would be getting advice from real, expert engineers who specialize specifically in AWS cost optimization. There was no layer of abstraction between analyst and expert. In traditional third party cost optimization engagements, a less-than-technical analyst might work directly with a client and then consult engineers for their insight. With Duckbill, there was no intermediary and no high-level abstraction. Duckbill would work directly with Epsilon and provide tangible, actionable insights. 

“We saw a significantly different perspective than other providers when engaging with The Duckbill Group,” said LeBaron  

Spotting Cost Centers within AWS

Epsilon had their own processes for finding anomalies in their AWS usage. They examine cost through a narrow, client-focused lens and a broader, company focused lens. 

Looking through the client-focused lens, Epsilon runs mirrored deployments of similar use cases. With their customer’s data on hand, they know that Client A runs a similar use case to Client B. So, the two customers’ cost per customer transaction should be comparable. If it’s not, Epsilon can flag that there might be an anomaly, and can go investigate to see why one use case costs more. Then, they’ll tune that anomalous use case accordingly to get it back within an expected price point. 

Looking through a business-focused lens, Epsilon monitors their AWS spend across their various product suites in the cloud and finds patterns in how they bring data in, store data, and move data. By observing the trends in their behavior, they can see where different APIs, or other costly services might be driving up their AWS bill and ask, “Are we being effective?” or “Why is the spend this high?” 

Observing these granular, and thousand-foot patterns help Epsilon set their own internal benchmarks for AWS spend. Duckbill helped them find areas that they can reduce that spend. 

Duckbill’s Observations and Impact

Duckbill didn’t just map out how Epsilon was using AWS, they mapped out how Epsilon could achieve key business objectives by streamlining their AWS usage. Duckbill engineers helped advise Epsilon on how they could store, move, and share data between systems, and pointed out which levers Epsilon could pull to save on AWS cost. 

One of the first things that Duckbill found when examining Epsilon’s AWS usage with a fine-tooth comb was data transfer costs between two different regions of AWS Athena. Duckbill identified the issue, brought it up in an initial meeting with Epsilon, found the stakeholder whose product was related to that Athena service, and Epsilon was able to make a change. 

“Right away, we were able to stop the flood of dollars going out the door related to that Athena service. We started getting that and other items extracted right away and we were able to take advantage of those savings,” said LeBaron. “Within the first one or two things that Duckbill found we more than paid for the cost of the engagement out the door and that was in a couple of weeks at most.” 

Duckbill validated what Epsilon was already doing well, and examined new, high profile initiatives Epsilon had planned. For example, while AWS offers managed databases like RDS, Epsilon thought that creating their own database might be the right business move, but they wanted a second, expert opinion. Duckbill agreed, helping Epsilon launch that initiative with more confidence. 

After several discovery meetings with Duckbill, a few project-focused meetings, and a thorough analysis of Epsilon’s AWS bill and architecture, Duckbill pointed out to Epsilon exactly where they could unlock vehicles to reduce cost reduction. These findings were shared across DevOps, IT, and engineering. With the cross-functional teams in the loop, it would be easier to make the changes that could help Epsilon save on AWS costs. 

“We weren’t asked to install anything or sift through unnecessary details about the solutions,” said LeBaron. “There was a very low level of effort for our stakeholder community to come in and engage with Duckbill and it still brought very meaningful results on the backside.” 

Those results speak for themselves. Duckbill initially estimated their findings would save Epsilon $3.5 million annually. Epsilon is already on track to hit that goal, and 12-15 individual areas Duckbill targeted for cost savings have already exceeded their respective savings estimates. 

Now, Epsilon can build with more budget, and more confidence. Instead of wondering how a change to their AWS infrastructure will affect their bill, they know in advance what the cost will be. Seeing engineering actions and how they’re coupled with AWS spend is a tremendous advantage for Epsilon. 

As Epsilon moves further into new architecture and transitions from EC2 to adopt more serverless infrastructure like Lambda, Jason LeBaron is keeping Duckbill’s number closeby.  

We’ll use Duckbill more to do a health check and help us out in the future. It’s good to have an expert like Duckbill come in and help us out. We’re confident we can extract cost and architectural value by engaging with them in the future.”

Jason LeBaron, VP of Solutions Architecture