AppNexus Abides As Mobile And Direct Deals Take Over

Brian O'Kelley_AppNexusIf you made it to AppNexus’ New York Summit on Wednesday, you might have come away feeling that the programmatic frontier has been largely colonized — at least as pertains to traditional display — and the remaining work to be done is along the lines of tree clearing and swamp draining.

VC investment has slowed and the number of new startups has declined in each of the last two years, noted Jerry Neumann of Neu Venture Capital, citing CrunchBase. Meanwhile many big publishers already have programmatic sales channels in place, and some companies are backing away from building display ad tech to focus on partnerships.

This can be viewed as healthy or perhaps discouraging, depending on where you sit. For AppNexus, a scaled company with 550 employees, it’s probably a good thing. But there’s still work to be done, especially around mobile and “programmatic direct.” (TV is still a remote opportunity.)

The company was late to mobile, but claims rapid advances. Most recently it signed a supply deal to become the trading platform for Millennial Media. On the negotiated buying front, it announced general availability of a trading/chat interface called Tango. (As in, “It takes two to…”)

“We are harvesting the benefits of seeds we’ve planted for years,” President Michael Rubenstein told the audience. “The next inevitability is mobile and that’s where our investment has gone.”

After the event, AdExchanger sat down with CEO Brian O’Kelley for a deeper dive on the company and marketplace issues.

AdExchanger: You seem to be talking less about product, more about partnership. Is the industry nearing maturity in programmatic technology for mobile and display? 

BRIAN O’KELLEY: What we’ve done in mobile is nothing short of amazing. When we showed Microsoft our progress, they were stunned. They said they’d never seen a company build so much product, so fast and so well. From an outside perspective, being in a position a year later to do a deal like Millennial, where they would choose us to be their mobile exchange platform, when we had no mobile technology a year ago? It’s just amazing.

It’s not that we’re not doing stuff. Mobile is the stuff.

What is now enabled in mobile programmatic, that wasn’t available before? Are you bringing tracking, rich media, targeting? 

We’re starting simple. We’re just accessing the inventory. We only did the deal six weeks ago. But the engineering teams are doing exactly that – they’re going through the formats, and saying one, technically how do we make it work? Two, how do we expose it? We’re going to have better mobile rich media and expandable video capabilities than we do in display.

The amount of work we have on creative control, on auctioning, on price prediction, on publisher and brand controls, on inventory selection, on targeting, on data — our pipes are just so much bigger than anything that exists today. The volumes are just night-and-day different.

On some level, making effectively a $100 million investment to build a mature exchange platform just wouldn’t make a lot of sense if you weren’t a market leader.

In fact, if you look at Facebook – some of the pullback on FBX – I think they thought that exchange-enabling Facebook inventory and calling it an exchange meant, oh, this is a general-purpose exchange platform. It turns out there is a massive difference between making your own API-based auction a real-time auction and building the entire stack necessary to integrate hundreds of bidders in real time to handle all of the actual business rules.

We’re really a business rules engine, right? Microsoft says, “We will not sell brand campaigns at lower than a $3 CPM in the US unless it’s a preferred partner or it’s AT&T, who we have a special deal with, and it’s nonexpandable.” That’s a real complex set of rules and logic to manage.

In what way is Facebook “pulling back?”

Have you heard them talk a lot about exchanges in the last 12 months? Eighteen months ago it was like, “We’re so smart. Can you believe how easy this was?” But the entire leadership team there – I mean, they bought Atlas, they did FBX, they tried a mobile ad network, and then whoomph, they disappeared. From my perspective. You don’t see them buying the rest of the pieces necessary to make Atlas a viable ad-tech platform. It’s not a criticism. It’s just a reality. If you’re Facebook and you want to grow your revenue from $7 billion to $11 billion in the next two years, what’s the shortest path?

What they did is, they did a deal with Google. Google agrees to properly attribute in DoubleClick For Advertisers, and we will let Invite buy from FBX. The reason they did the Atlas deal was so they wouldn’t have to do that deal with Google.

I don’t understand how Google wasn’t attributing correctly.

I don’t either, but it was the reason they bought Atlas. Gokul [Rajaram] said, “We have to buy Atlas because we don’t trust Google to give us proper credit. We can’t be reliant on DFA.” Their whole thing was, “We’re going to fight a war against Google. We’re going to make Atlas the measurement standard for the Internet. And we’re going to make FBX the ad-tech platform.” I’m just saying, it’s a complete capitulation.

You’re also seeing all the Facebook partners either change their business models or go out of business.

All of this is a playbook. Before, it was, “We’re going to be an ecosystem player.” Now they’re saying, “We want to be like Google.” Which I completely respect and understand, but it is a very significant shift in how they’re operating.

It’s not dissimilar to the conversation we’re having about Millennial. The amount of tech and partnership and enterprise muscle that’s necessary to run a real ad-tech business is extraordinary. We have a 12-person team just doing integration, plus the account management plus the engineers. Why would you do that unless you’re committed and all in to trying to be effectively the hub of a commodity market.

Would you put native ad formats in the “mobile rich media” bucket?

If it’s truly native, the answer is no. Truly native would be search. Social platforms…that’s different. Would we allow you to do a News Feed ad in Facebook, the answer is yes. And Twitter, what’s our role in the Twitter ecosystem? Would we work with MoPub? Sure. But what would be the value-add for me to put a sponsored Tweet up in real time.

Isn’t it the same value as putting up a display ad in real time?

I would argue no. Because display ads are standardized. They’re meant for distribution to many platforms.

Couldn’t a Twitter Promoted Tweet be standardized, if they do a real integration with MoPub? Wouldn’t they effectively be pushing native formats out to other properties?

Sponsoring a Tweet on anything that’s not Twitter is still a Tweet. If you’re saying I can run an ad that looks like a sponsored Tweet off-Twitter — it’s not a native ad anymore. Now it’s a scaled, standardized display unit. It might have a text ad inside of it. That’s what Google realized with AdSense.  If Twitter standardized it, it’s just a text format. Which is cool. I don’t have a problem with it. It’s not native.

With Tango, you’ve added one more thing that seems to make a technology like an SSP unnecessary.

It doesn’t make them unnecessary. It just forces them into being media businesses – effectively networks. Go to Rubicon Project’s website and you’ll see it has an advertiser cloud and a publisher cloud. The whole point of the company was to be publisher-only. Why are they giving publishers and advertisers equal status?

What about PubMatic? They’re a little different.

PubMatic’s dead, because they don’t do that. There’s not enough money in the pure publisher business. Rubicon says, “If you work with us we’ll give you more money than PubMatic will, because we have direct relationships with advertisers.” Which is what said eight years ago. And by the way, AOL is trying to run a marketplace play around ad tech. Xaxis is trying to run it through 24/7. Everybody is running the same playbook. The problem is, the playbook doesn’t work if everybody runs it. Criteo runs the playbook. Everyone is an ad network. And if you’re not an ad network – like PubMatic isn’t – you’re dead. There’s no room right now, economically, to grow if you’re not an ad network.

How do you define ad network?

A media company that makes money by selling media to agencies and buying media from publishers. Why is Rubicon making price guarantees to publishers if it’s a tech platform? Because they need access to media, and they’re gaining advantage by getting more media than PubMatic. PubMatic won’t make guarantees because they can’t. They don’t have any margins.

It plays into our hand. We can walk into publishers and say, “Look, Rubicon’s great. They’ve got a lot of direct advertiser deals. But you need someone you can trust to manage your tech business, because Criteo and Rubicon are making the exact same pitch to you.”

So if it’s true that these guys are trying to get “first look” or a chunk of inventory, how do you make sure you’re first in line?

We don’t care. Our whole play is: As a publisher, you need a tech platform that helps you manage this complexity. That’s what Tango is. You build whatever relationships you need to. If you want to put Xaxis first and Accuen second, click “1” and “2.”

What’s going to happen to the trading desks?

Bottom line, each has to build a viable network business inside the holding company. We can all call them whatever we want. Either you take IOs and you effectively buy media from publishers, or you don’t. And who doesn’t, aside from us and maybe PubMatic?

Our focus has always been on the bigger companies, like Microsoft, that need a technology solution.

Any updates on the business?

We’re so deep into 2014 planning right now. One nice thing about not being public is we don’t have to talk about it. It’s impressive growth, and much faster than the rest of the market. But as we move down this path, it matters less to us than it does to others. Our side of the market is continuing to grow faster than the market as a whole. We don’t have to worry about quarterly renewals, we don’t have to convince some media buyers to give us more money.

How are CPMs trending in display?

I have no idea. I don’t care since we’re not a media company. It’s not one of the 50 KPIs that comes across my desk, since it’s not how we make money.

Ok, what are some of the 50 KPIs then?

I’m deeply concerned with: How are we doing with key customers? How is our optimization technology performing? How is our uptime?

I care about product release cycles, customer pipelines. It looks a lot more like a SaaS business. Ironically, we could be in a different market and most of the company would just shift. One of our clients said, “I’m shocked how little a lot of your employees  know about advertising.” I’m not shocked at all. We’re a technology company first, that happens to build trading systems for advertising.

Will you go public?

It’s neither yes or no. Our aspiration is to some day be a public company. But that requires a lot of work and preparation. There’s an argument to be made that we should be rushing out the door to take advantage of the hot market. The problem is I still have to run the company when it’s not a hot market. Having the hottest IPO in 2013 means very little in 2015.

What makes more sense? Let Tango come out and see how it goes. Let our mobile investment come to fruition. Really launch the right way in APAC. At some point it will be good business, when the business is predictable and scalable.

It’s just another financing round. Why would I want my investors to be irrational hedge funds?

Enjoying this content?

Sign up to be an AdExchanger Member today and get unlimited access to articles like this, plus proprietary data and research, conference discounts, on-demand access to event content, and more!

Join Today!


  1. Tim Ogilvie

    This is awesome. You couldn’t get Brian to say what he was really thinking?

  2. Great conversation here…

    I’d be curious to know how much of Appnexus revenue is based on percent of revenue vs. a flat ad serving fee.

    I think it’s mostly percent of revenue, from both a buyer AND a seller, so I don’t understand how AppNexus can fully say it’s not a network or a media company. If AppNexus was purely a straight CPM pricing like an ad server for all its customers then it would be technology only, SaaS even. But it does thrive off of getting inventory and buyers while taking a percent. Despite what the departments internally are named or whether or not there’s a “Publishers” section and an “Advertisers” section on the site, this is a network model.

    Because of this, I feel like AppNexus has to care about CPM.

    Example: “We can walk into publishers and say, “Look, Rubicon’s great. They’ve got a lot of direct advertiser deals. But you need someone you can trust to manage your tech business, because Criteo and Rubicon are making the exact same pitch to you.” – This is the same pitch that Rubicon and Criteo actually make about AppNexus but only the word “tech” is in there… and I bet sometimes they even throw that word in there in place of media all the time.

    HOWEVER, the big differentiating factor above in the pitch is that AppNexus will in fact *customize* the technology for companies in order to get them as a customer… to the point of full blown white labeling. A huge difference, one that has paved the way to success and locked in the clients that opt to customize. If you don’t customize the tech, or use them as your sole ad server though, AppNexus is another source of demand and supply in your chain that competes with everyone else.

  3. Great interview, and congrats to Brian and AppNexus for the continued success. An online/mobile ad network is a great business if you can do it at super-scale (like Google, the ultimate ad network), as is a tech platform, but most companies are stuck in the middle: both from a scale and a business model perspective. Brian’s just calling out what to most should be obvious by now.

  4. Yoav Arnstein

    Most eco-systems, unlike tech platforms, are complex universes that are influenced by systems and humans (yes, non programmatic creatures). It makes them difficult to anticipate.

    In our eco-system in particular, with so many private companies and a limited (but growing) amount of public companies, a great deal is unknown.

    In plain words, there is always s..t we dont know.

    As such, i believe that all predictions, assertions and opinions should be accompanied by an essence of humility.