Joe Zawadzki is CEO of MediaMath, a demand-side platform technology company.
As part of its "State of..." series of articles with industry executives, AdExchanger.com sat down with Zawadzki to discuss his company, his views on the space, and the state of MediaMath today.
Click below or scroll down for more:
- Today's M&A
- The Past Year for MediaMath
- The Challenge of Programmatic Buying For The Marketer
- Is MediaMath a DSP?
- The Meaning Of "Trading" In Digital
- If I Were A Publisher...
JZ: I think the market has reached a crescendo. Like every market, there are some natural periods of expansion and consolidation. People are now starting to see the complexity out there and some of the smaller, independent companies might be thinking that market forces are asking them to either be platforms and aggregators, or be aggregated.
There are too many places to spend dollars and too many sales folks showing up in agencies for us to credibly vet all of them. Buyers are saying, "I want to double down on the bets with those that I trust and work with. I expect them to close the gap on feature sets with new innovations."
Meanwhile, smaller companies are wondering, "Hey, am I a company or am I a feature?" The more people that do that, the more M&A activity there is.
Business has been moving forward on all fronts. People‑wise, we're up to 150 now. One big change in the past year has been in our geographic scope. We have offices in Boston, Chicago, San Francisco and L.A. Two of MediaMath's co‑founders started up the London office in February, and now there are 11 people and "north" of a couple million dollars in spending per month in Europe.
We've also put a data center in Hong Kong, have a Japanese trading desk through Hakahodo, two trading desks in Argentina and trading desks launched in Mexico. You're starting to see a global footprint. The plan is to open up a Sydney office by the end of the year.
In the U.S., we still have a mix of clients that aspire towards self‑service, but continue to lean on our team to do the more complex trading. Internationally, it is 95 percent self‑service.
Also in the past year, clients are getting trained much more quickly and that leads to a scalable model where they can come in on a Monday, and then by Tuesday, they have "live campaigns." The speed totally changes clients' staffing requirements.
Another big change for us is the scope of the platform. We started in the RTB display world and exchanges because of the big supply-demand imbalance. Fast forward to now, and we’re buying through a private marketplace and the onboarding of a lot of premium supply. The platform's become the buying platform for all digital, and less search. That's obviously consistent with the roadmap and vision of buyers that want to have a "Bloomberg terminal" - a normalized interface workflow with consistent data, consistent dashboard, and global reach of frequency across all of channels.
Has the conversation around online behavioral advertising and privacy inhibited the business that you're in?
Well, the conversation probably creates a little bit of reticence for people who haven't tried it.
Do you think marketers are a little hesitant at getting into this business right now?
I don't think so. I think behavioral targeting, specifically, like the targeting of third‑party audience profiles is a tool. It's not the business of online advertising. Some companies are heavily reliant on it. They buy audience segments and test against them.
Since online marketing is about having a huge amount of reach and a huge amount of attribution ability, you can reach segments of users on sites by using audience profiles, authorization, or though going out and scoring valuable impressions. Only one of those things is heavily reliant on BT.
"People" is a big challenge. They actually had this Times article recently about it where it's really hard for organizations to assemble the four things that are needed to do marketing and certainly digital marketing well…
An employee would need to know the ins and outs of marketing, and specifically, accountable marketing - the ability to connect how they affected results. The notion that "T.V. works and we're not going to measure it" is dumb so we need data‑driven marketers, who can apply their skills to business problems.
Also, an employee needs to understand technology and, most likely, they need to build technology - you need to know quant to do data‑driven marketing well and manage something like 17 billion impressions a day. You're not managing this data with Excel or Access. If you're analyzing billions of impressions a day, you need a machine.
Finally, in order to make the best use out of the inventory, you need to literally figure out what to pay for each of those impressions and deliver the results that you want - you need to have a huge amount of domain expertise.
And all that is on the marketer's side?
That is a responsibility of the ecosystem to provide. If you want a good ad tech company to build a good product, you need all four of those qualities. With domain expertise, you need to know how it all gets done - and know the booby traps. It's a nuanced and arcane industry.
And it's a big proposition of the demand-side platform (DSP) to ask, "How do you integrate those things and put them into a technology or a platform that you could then license to somebody?"
Do you run up against culture as an issue on the marketer's side or is it about having the right skill set?
There is a cultural bias. I think most of it is our fault. You just don’t "get involved" in advertising technology. You didn't do it unless you were fundamentally geeky. So let's call a business-like Quant and Tech the next rev of what you need,
And if you're geeky, and you love knowing how things work, then you actually like understanding how the sausage is made. A lot of marketers, for very good reasons, don't actually care how the sausage is made. .
A CMO that walks into a brand, has to figure out an overall landscape. The job is part product, part merchandising, part fulfillment and part advertising. And they are doing the job in an organization quickly because their tenure is short. To think that they also have to get in and understand how pixel‑based targeting works, and what ad verification is, it's asking for them to say, "No. I have bigger things to worry about than that." There’s clearly cultural resistance. If you don't know it, and it seems both annoying and complex, people tend to resist it. However, both sides, if you will, need to take one step forward.
We need to stop selling speeds and feeds. We need to start selling benefits and talking about how this is going to work for the marketer. Marketers need to start thinking, "I have to put a system in place to make all this stuff make sense."
I'm actually fine with it. But the definition of a DSP has come to be more expansive. It's not your grandma's DSP. There is bid management on exchanges, and we're not that. We are truly a demand side platform. We are the entry point for demand to normalize the buying process, the optimization process, the analytics process across a marketing channel. If you think about how search gets bought, you pick a search management platform and it manages your demand. So I think demand side platform will become the entry point to all demand, not simply RTB on exchanges.
Today, people are using the term "trading" for something that isn't widely prevalent right now in advertising. Right now, what’s prevalent is automated, analytics‑driven buying so you're buying on behalf of a brand and delivering against their goals. Very few people are buying and selling. There's no derivatives market - there's not really a futures market in media. Not that there won't be in the future, but it's pretty new right now.
Do we get to a financial markets model in advertising, in the future? How far away is that?
There's one big fundamental difference in financial markets versus media markets. Financial markets are bigger. Bigger markets tend to create innovation faster for obvious reasons, since there's more demand.
But, there is also an intrinsic value of a stock. It's the price in a moment of time. There is no intrinsic value of an impression, and it's also very perishable. Therefore, every advertiser will value the impression differently - for different offers, and depending on their metrics. They'll value the impression differently depending on the products or the call‑to‑action on the creative.
So different verticals will value things differently, Capital One will value it differently than an American Express. There are hundreds of price points for any impression, depending on who's looking at it. So it makes it harder to trade derivatives.
You could do interesting things like predict the average value of a Yahoo impression over time. But then you need a pretty sophisticated market player that really just wants to gamble on market prices, not necessarily providing a ton of value.
I think there is the ability to do true up‑fronts – as in lockup inventory - since a buyer may think inventory is mispriced or because they think they can add more value to the inventory than the direct sales force can. That’s an interesting opportunity as well as the opportunity around offseting or offloading market risk.
If you went to a marketer and said, "I am going to guarantee your orders through the digital channel for the next year," and then you take the market risk in terms of making that happen, and profit. Well, that's almost a commodities market or option market. Then, I think you have some interesting financial markets metaphors.
I know it sounds like it's self‑serving, but if you put me in charge of a publisher’s inventory, the "Class 1", direct‑sold - effectively "upfront" - doesn't hold up anymore because there is such a groundswell of demand for this new way of buying.
Historically, programmatic buying has equaled remnant and traditional buying has equaled premium. The benefits are so acute that the buyers will demand all media be bought programmatically. That means you've got to figure out how to make your inventory available through programmatic channels where you still capture the upside of "premium‑ness." As a publisher, you've got to find a way to tranche your inventory through programmatic channels, and be able to say, "Full transparency costs this much, partial transparency costs this much, and opacity gets you this much of a discount." Data use - the ability to layer in third party data and the ability to layer out the second party data to be used in other places - all will come with different prices.
The "premium‑ness" of the advertiser and my trust in its adherence to a contract will earn a discount or earn access to inventory that otherwise wouldn't be available to the open market. This is the ability to influence the supply mix with big enough advertisers.
Also, publishers don't have a fixed inventory. If they produce new content for a specific advertiser-consumption, advertisers should pay for that because they've literally just expanded their supply chain.
But, it all goes back to the publisher team. You need somebody that's looking at this stuff and saying, "What unmet parts of the market exist? How do I understand more about the demand, such that I can have more influence over the supply?" As a publisher, if I can do all of that, I can suddenly take this big pool of programmatic remnant, and suddenly create tiers that are not one dollar, but two dollars, three dollars, six dollars and nine dollars.
The real challenge is, it doesn't happen slowly. There will be a couple of publishers that just smash through this and end up being the object lesson for folks that are tentatively doing this.
Another thing - folks also forget that a healthy publisher ecosystem leads to happier buyers. It is not an antagonistic relationship. If publishers aren't thriving, advertisers don't have places to buy. So you need to solve for both sides of the equation.
I've been in the industry now 13 years and with regularity, the pendulum swings from buy side to sell side. This year and last has been the furthest out on the demand side that we've gone. But, it's started shifting back again. It’s all very natural.