- Why Change?
As a disruptive company in the highly competitive event ticketing market — SeatGeek delivers superior customer experiences and innovative features, such as DealScoreTM, at a competitive price. As a result, ensuring that costs stay under control is essential.
Like many software companies, one of their largest drivers of operating cost is their AWS bill. In 2020, the infrastructure engineering team, led by Michael Weinberg, knew that they needed better visibility into their AWS spend.
On one hand, the entire live events industry had recently come to a screeching halt, which meant they needed to ensure they were spending wisely while the industry recovered.
On the other hand, they had seen incredible growth over the past few years in SeatGeek Enterprise — a powerful stack of primary ticketing services utilized by big name organizations from the Dallas Cowboys to LW Theatres.
On the consumer ticketing side, they needed to quickly understand where they could optimize and redistribute spend to other areas of the business. On the enterprise side, they wanted to ensure they were maximizing revenue and pricing appropriately for large accounts.
According to Michael, their homegrown cost reporting — which was built with a combination of the AWS native tools and their infrastructure monitoring tool — wasn’t cutting it. While it was adequate for reviewing high level trends, they lacked granularly or connection to the products they were building.
Michael would often get questions about cost from other teams, who wanted to report on financials, understand changes, or make decisions about products or pricing.
“Every time the spend conversation would come up, I’d probably spend six to eight hours over the course of a week fiddling with things, digging into data, and taking different slices. With Cost Explorer, I’d see the change, but it wouldn’t answer the question of ‘why does this seem different?’ — so I’d have to figure it out.”
This was particularly challenging for SeatGeek Enterprise. They had many big name customers, but the costs to support these customers were opaque to the teams responsible for managing the deals.
“Every time the spend conversation would come up, I’d probably spend six to eight hours over the course of a week fiddling with things, digging into data, and taking different slices.”
Michael and his team implemented CloudZero, a cloud cost intelligence platform that delivers relevant cost data to engineering teams about the products and features they’re building.
Immediately, they gained granular visibility into their spend — as well the layers below their spend categories, which has previously been obscured.
“We have a sizable monthly spend and compute is about 70% of it — so it easily masks everything else — and even major changes can be hard to spot. With CloudZero I can dig into my bill and easily understand exactly what we’re spending and why,” said Michael.
This is a CloudZero image with sample data, not SeatGeek's spend.
With CloudZero, SeatGeek was also able to produce “multi-dimensional slices of cost” that were relevant to their business. For example, they aligned cost to product features, normalized unit cost, and generated a cost per customer report. Because of CloudZero’s machine learning, they did this without manual tagging exercises.
This is a CloudZero image with sample data, not SeatGeek's spend.
For SeatGeek, the biggest cost challenge they had was measuring normalized unit cost. They had worked with AWS and looked at other vendors, but had yet to be successful. SeatGeek runs their own container architecture, which overlays AWS services, and obfuscates their costs.
As a result, they could see high-level cost trends, but not how much it cost to run their product features.
“We wanted to know ‘What is the cost impact of not optimizing API memory usage?’ or ‘This is how much it costs us when this thing has crashes and this is how much it costs us to have the CPU overrun.’”
Now, they can measure the cost of individual applications, product features, and more — giving them essential decision making context.
“It’s really beneficial to us to be able to just create the reports that people actually would really like, when in the past, it's just been completely opaque,” said Michael.
For SeatGeek Enterprise, they can now also understand and communicate how much each of their individual customers costs.
“For our financial planning and revenue teams, they get a report that says this is how much it costs to service the Cavs — or a little theater. They can approach the sales or renewal process a lot more intelligently. Having that information without a lot of back and forth is huge for them.”
“[Our revenue team] can approach the sales or renewal process a lot more intelligently. Having that information without a lot of back and forth is huge for them.”
With CloudZero, Michael no longer has to spend days digging into costs. CloudZero helps him understand his unit economics, so he can quickly answer questions from other teams and executives.
Because CloudZero automatically correlates cost with resources and engineering activity, Michael saves time exploring spend, answering questions, and addressing cost issues.
“It’s really nice to be able to just send someone a link to a view to show them what I’m talking about — then they have the context right there to investigate with a simple workflow. You don’t get that level of efficiency with AWS or the other ‘fancy’ cost tools.”
For Michael, CloudZero’s value isn’t just about saving time or a little money — it’s about controlling unit cost. Drilling down from a cost to its associated resources gives the SeatGeek engineering team meaningful context about whether it’s a one-off line item or a repeating, impactful part of unit cost.
“With other costs tools, we could get a very broad view of cost or line-item accounting — but we couldn’t traverse between the micro and macro levels. The level of detail CloudZero gives — plus the ability to correlate those levels with a couple clicks — is really helpful when we’re trying to make a decision about how much time we should put into a problem. Because, for example, $7 might not be that significant, but we need to know if it’s $7 times 1,000 volumes or hosts.”
CloudZero also helps them to quantify cost and prioritize optimization — so they’re ensuring engineering resources are only spent on the most impactful projects.
“We’re definitely more cognizant of how a decision that might feel like a trivial implementation detail can have a huge impact on something much larger, like whether we can hire.”
“The level of detail CloudZero gives — plus the ability to correlate those levels with a couple clicks — is really helpful when we’re trying to make a decision about how much time we should put into a problem.”
With costs put into context, the engineers at SeatGeek have become more aware of the impact of their decisions.
“We’re definitely more cognizant of how a decision that might feel like a trivial implementation detail can have a huge impact on something much larger, like whether we can hire,” said Michael.
Engineers have identified quick changes they can make to optimize, then get feedback immediately in CloudZero that shows them how much they saved. “The positive feedback loop is really motivating,” explained Michael. CloudZero’s cost intelligence data also enables them to have productive conversations about the effort vs. outcome of optimizations before they make them.
Michael sums it up, “We want to have the best inventory at the best price. The more we understand our cost to sell and deliver each ticket — the better we can do on price. That’s good for our business.”