MediaMath Video

MediaMath, a New York City startup with an idea to help ad buyers capitalize on the rapidly evolving marketplace by providing services and tools for ad agencies – the most active “demand side” players – to identify, bid on and buy just those impressions most likely to yield the results they sought for their clients. Implementing the idea in 2007, MediaMath created a hot market segment now known as demand side platforms, or DSP. MediaMath needed serious analytical power to allow sophisticated buyers to harness and channel data to drive optimum performance for any advertiser, campaign or marketing objective, using any combination of data inputs. Ultimately, the Netezza Platform provided MediaMath the ability to consume massive amounts of data from disparate sources and come up with decisions in real-time. In this video you will hear from Roland Cozzolino (CTO) and Tom Craig (VP Data & Information Strategy) about MediaMath's analytical objectives as a leading Digital Media firm and the role Netezza plays in their business strategy.

Video Transcript

The MediaMath TerminalOne platform is a combination of technologies that allow people to trade digital advertising. By trading digital advertising, I mean the ability to purchase the opportunities while they occur, take that data, analyze that data, perform yield on those buys, and do this over and over and over again – all wrapped up in a single interface that a user can easily work with to effectively buy spot ads.

Tom Craig, VP Data & Information Strategy, MediaMath:
Large scale data and analytics is at the core of MediaMath. It is the lifeblood of this company and it’s the lifeblood of the entire ecosystem of advertising.

There are a couple of critical elements that enable MediaMath as a company to buy the impressions that we want – essentially the impressions that are going to succeed. It’s basically two fold. The first thing is our ability to collect all the data, store it and analyze it. This is where Netezza comes into play and fits perfectly. Without that, we’d have nothing. That being said, once you have all that data and you know how to analyze and optimize against it, the second portion is our ability to write code that runs fast enough to buy those impressions across n segments where n could be anywhere from 20 to 50. Usually within, for example, the RTB [real-time bidding] space, you have 30 milliseconds to respond – of which the network takes up 25 milliseconds – giving me five milliseconds to figure out, ‘Is this worth it for me?’ Multiply that by the number of campaigns that are running, the number of advertisers that are running on you, and you’ve got a massive, massive problem. What we’ve done here is spent all of our time utilizing products like Netezza, utilizing our background in the industry with respect to advertising, and building out a system that can execute these things across all of these segments as fast as possible.

The most critical thing in our ability to act in a real-time environment is speed – the ability to consume massive amounts of data from disparate sources and come up with decisions that you need to make at that point in time.

MediaMath did a rather extensive ROI evaluation on all the products that essentially claim to compete in the same space as Netezza.

From the first day, we were treated as the biggest client that Netezza ever had and it was fantastic.

Some of the competitors of course have good products. Nothing did exactly what Netezza did. Nothing performed as well as Netezza did. The support Netezza gave and gives today is unparalleled, in my opinion, to almost any technology we’ve ever purchased. So while we ran all the numbers, while we brought terabytes of data to all the competitors, Netezza won hands down on every single category we looked at – which is execution, which is support, which is quality product, which is ease of use.

Things that looked impossible to us prior to the implementation are now second nature. Processes that would take days to finish or maybe even the harder ones that wouldn’t even finish are run hourly now.

You hear people talk about things like using Hadoop, and you know, clusters of servers and all these types of technology that you have to build. The reality is it’s not practical. Netezza has made for us the seemingly impossible, completely possible and practical. I don’t have to worry about, ‘how am I going to mine 12 terabytes of data in less than an hour?’ I have yet to bring something to Netezza that it couldn’t handle.