George John is CEO of Rocket Fuel, an online advertising technology company.
As part of its “State of…” series of articles with industry executives, AdExchanger.com spoke with John to discuss his company, his views on the space, and the state of Rocket Fuel today.
Click below or scroll down for more:
- RTB Impact On Display
- The Ad Network & Company Positioning
- Revenue, Momentum
- Trends Ahead for 2012 and 2013
- Mobile, Social And Rocket Fuel
GJ: It’s much better than I thought. If you consider the layers of sophistication around leveraging exchanges and liquid inventory, where “1.0” was the original Right Media Exchange, and you’d log in and define your targeting rules and your bids, then “2.0” was real-time bidding – but it doesn’t add anything to that style of buying.
And, if you thought in terms of rules like “I love this BlueKai segment so I want to pay $3.50 CPM whenever I see someone who seems like they want to buy a luxury car,” then real-time bidding doesn’t add a ton. It lets you build your own computers and manage your own global frequency counts across multiple exchanges and all that. But if you just think in terms of rules, it adds a bit of efficiency, but not that much.
What we’ve done is plugged into the major [RTB] sources for inventory that give us 10 billion impressions a day and that we evaluate through four data centers across the US and Europe – and we’re building out in Asia. Every impression gets fully scored by AI (artificial intelligence) models that predict response. I know everybody says this.
For a while, everybody thought that it’s all about just aggregating as much supply of ad space as you can. And then, it was all about getting as much data as you could. When I was at Yahoo, we suffered from this disease of thinking that more is better. In fact, it was the mission statement of the Yahoo data group back in 2005 to 2008, to build the world’s largest and most actionable database.
[Former Yahoo! CEO] Terry Semel once asked a good question of the leader of that group. “What’s so great about it being big? Most actionable I can understand, but does it have to be big?” We feel that RTB is a critical enabler because only by looking at every impression one at a time can you have an A.I. model saying, “This one gives me the chance of response at 1.03797 percent so I’m going to bid $20 CPM on this one impression.”
Before, if you had intelligence, there was no way to express it in the way ad servers and pre‑RTB exchanges worked.
It’s funny. I think of this old joke people used to say about IBM, which was that you could tell IBM didn’t invent the car because if they had, they would have put a saddle, stirrups and reins on it to be compatible with the horse. But it’s what we did here, at least on the way we addressed the market. We built a car, or even a cool spaceship, but we still put the saddle, stirrups and reins on it because that’s just the way media buyers are used to operating. They entertain proposals from publishers and ad networks.
What we love about this whole ecosystem is just that sometimes it feels a bit random getting in the door and getting attention. But once you’re in, it’s quite rational. [Agencies] have their DART or Atlas reports and they can look side‑by‑side: “here’s how Rocket Fuel and Yahoo did, and other ad networks” etc. It makes them want to give us more money -so we like that model. The DSP as a category would be the alternate because what we’ve built from a technology perspective is more than what they have as an ad network. We have none of the supply‑side of an ad network. We have no publishers that we have any commitments to.
Every now and then, we’ll do a spot buy on a publisher and try to figure something out for something inaccessible through exchanges. But 99 percent of what we buy is through exchanges and it’s all bought through this A.I. engine. In terms of buying everything off of exchanges, we look more like a DSP. In terms of using A.I., I’m not sure that we look like any of them to the extent we rely on it.
I’ve seen other people in our industry, like Brian O’Kelley, who just look up at the ceiling and roll his eyes when people ask him what category he’s in. I share his view. I don’t know what to call myself. But if someone has ads to run and an objective, I would love to help them out..
I’ve never been too shy about that. We did about $44.5 million last year, which was a little under three times the year before. This is the top‑line revenue. So far, I’ll say we’re significantly ahead of our plan for this year. My sense is that we deserve at least a bit of credit for that, but also the space is growing. It’s weird that it’s taken this long for the industry to get credit for itself, but you’ve seen some major advertisers talking about seeking efficiency with their ad spend, like Proctor & Gamble’s talk about shaving off ‑‑ I believe the number quoted there was $6 billion ‑‑ in ad spend by getting more efficient and leveraging digital more intelligently.
We’re benefiting from a lot of tailwinds, but within particular customers, like how we worked with one large telco company. Just as color, the Q2 buy from them last year was $100,000, Q3 was $500,000, and Q4 was a $1 million. We’re aggressive about achieving results for advertisers. We’re getting scaled up pretty fast. We’ve been strong on the verticals with automotive as our top vertical right now. Automotive advertisers are regimented about the way they measure post‑ad website activity, so they’ve got certain key performance indicators (KPIs) for measuring how much it’s been worth to them to get someone to download their brochure or sign up for a test drive.
We have 15 customers right now that have done over $1 million with us. They span automotive, education, travel, telco, financial services, retail, health and consumer packaged goods.
Basically, what we find is that the closer you can get as an advertiser to helping us measure your objectives, the closer you can get to sharing with your ad partners what their value is to you – the better the results get for you
I can credit some of our growth to us and some to the industry. Advertisers are just getting more quantitative. It comes back to CFOs, CEOs, boards of directors and attentive CMOs. They want to understand better what happened with their money. There’s a danger of “analysis paralysis,” but excellent business people can figure this stuff out and build, at least, some basic media mix models that can help advise them on how to spend their money.
Some of the companies that are doing that, they’re the ones that are shifting money into digital because they’re realizing that’s what’s driving a significant portion of their results. There are some great comScore studies, Nielsen and others about how the synergy between the two seems to have that effect, versus spending it all in either medium alone. As Rocket Fuel has evolved, we’ve become more strategic to our advertisers because we’ve begun to take on a significant part, if not 100 percent, of their display budget.
As we become more strategic to our customers, we’re participating more in the conversations that do talk about how they’ve been trying to figure out media mix optimization. What’s been gratifying is they’re looking to us to help with that. We have some attribution experts on staff and we do things internally so that, even as we’re optimizing a campaign ourselves, and we’re not getting misled just by the last click. It’s interesting because when we first started Rocket Fuel, we were a simple display ad network. Whenever I would meet with the CMO, the response in 2008 was always, “This isn’t interesting to me. I’ve got a guy who’s got a guy, who has an agency that manages that for me.”
Then, I would ask them to forget that. I’d say that I’m a smart guy who just managed to convince a venture capitalist to give me $10 million, and I’m going to build something for him and them. The universal response was interesting. They all know that whatever they’re doing right now, it can’t be right. It can’t be optimal. They’re thirsty for more data‑driven help. There are a lot of factors that are coming together all at once that are allowing this to happen for them.
Are you finding that you’re having to build out your service side more?
Right now, we’ve gotten some customers that basically want us to be their agency. We want to get an IT sense on their strategy, but it’s the hardest thing to do: to build a successful agency. You have to scale yourself up with smart people that can have these strategic discussions with customers. We have a number of them here, but if I look at how our business scales out for another two years, it’s going to be a real chore to staff up that team successfully. I would say it’s something that we’re letting customers guide us on right now. What we really want to be is a software company.
We’ll want to have a leveraged consulting team that doesn’t do everything, but can work with an agency to help the right things happen with a customer, and can advise a customer on the right things to do. I think that’s doable. The startups I’ve been at before, they managed to maintain the focus around software with the appropriate level of responsibility and helping the customers use it well, such as Salesforce and Epiphany. If we grow that way, it is a highly leveraged solutions team that works with an agency and a client to get things done.
Do you think agencies, today, are well‑equipped for the opportunity that Rocket Fuel, and perhaps others like you, are offering? If not, where could they improve?
There are a lot of agencies that do a fantastic job. But, I’ll share a message I received from an agency a month ago now. They actually sent our account manager a lovely note that was thanking us for our performance and gave us an anonymous table of results, showing which lines were us, and which lines were other networks or publishers.
We were number one by an incredibly wide margin for doing online sales of a consumer technology item. We had a $50 CPA (cost per action), and they had two partners on the plan at the bottom of that list with the same budget as us, delivering a $500 cost per action. They were telling us we did a good job. What was irritating and mystifying was – How could that email not have ended by saying, “So, therefore, we’re taking half the budget away from ‘bad guys’ number four and five, and giving it to you?” I feel like, in some cases, the agency is just not as attentive as an advertiser would be, directly, if they were seeing these numbers.
Sometimes, that’s excellent because for all I know, the bad advertiser number five did some amazing, innovative things, and they did need some time to optimize. Generally, agencies sometimes don’t seem to care as much as I know the client would. It’s useful to have them as a buffer, but in terms of what agencies could do to serve clients better, I feel like there’s a gap in training.
I know many of them use some of the institutions in our industry to get the new hires up to speed, but there’s some laws of physics that exist now, based on the way exchanges work and the way real‑time bidding gives a lot of players access to this massive ocean of inventory. In the old days, maybe you worked for the partner and you couldn’t scale them up so there was no point in trying. Now, for folks like us, we can scale far more.
Where this shows up as a larger hindrance on performance, or at least as impactful as the media, is the creative side. As far as we’ve seen, every agency still does creative as if it were a magazine. They had to have the creative done on Sunday because the magazines are going to print on Tuesday, and once it’s done, you have no chance to make it better. Of course, in digital, it’s a real‑time medium. You could come up with three sketches, storyboards or early versions of one, and run it for a week ‑‑ or even a day ‑‑ and get good statistics that tell you which version you ought to develop into the fully‑polished and possibly the flash ad. I find that training folks to understand what the current opportunities in digital are, particularly enabled by real‑time, helps. That notion of real‑time, figuring out everywhere you can leverage it, is the gap right now.
First, mobile is a funny label for a lot of things. It could mean people browsing the web on their iPads, Android tablets or a smartphone; or seeing little mobile ad units on the bottom of Angry Birds, or even in SMS and on feature phones. It’s clear that Mobile is significant.
We’ve looked from an analytics perspective by device types to see how each device is affecting people’s response rates. And we’ve seen that mobile corresponds more or less to a common sense notion that a larger format tablet device is more for the couch and reading. We see nice response rates there. Smaller format tablets, we see lower response rates. There’s less space to scroll around and see the ad.
Then, phones have been among the poorest because the user’s modality is a bit more on‑the‑go, for instance, trying to look up some address while on the subway or something like that. If you take an index of one for the iPhones, you get twice the response rate on the iPod Touch, which is weird because it’s the same device, just different usage modality. And then, it’s about twice that again for an iPad.
In terms of the demand that we’ve seen, we don’t work a lot with the more incestuous type of mobile advertising where it’s game producer A trying to run ads in game producer B’s apps, to download game producer A’s apps or something like that. It’s more the top brand names that we would be working with. For them, I can say they’re still trying to understand what mobile represents to them in terms of an efficient channel of engaging and reaching consumers. Also when I say mobile, I’m meaning more of the Mobile Marketing Association (MMA) ad units or other custom apps.
A [display] banner or video ad – those are part of mainstream planning. Among, video, social and mobile, video has the strongest uptake because it’s closest to what advertisers already understand. It’s social and mobile that have special teams that are trying to figure them out. We see mobile as the farthest out.
And social for Rocket Fuel? Thoughts?
We look at social as having three components.
Social data – such as your Klout score or Rapleaf has data suggesting how well‑connected people are, etc.
Next, there’s social inventory and actually running ads on Facebook, which we do.
A quick sidebar – my experience at Yahoo! prior to coming to Rocket Fuel, was motivated by monetizing social inventory. Yahoo! had this annoying problem, which was that Yahoo! Mail accounted for something like 40 percent of all display impressions, back in its heyday of 2005-2006. The trouble was, how do you make any money off that? If you just sell it as Yahoo! Mail, some random person reading their mail, how much would you pay?
That was why Yahoo! Invested in behavioral targeting, which [unlocked] somebody who just spent the last hour on Yahoo Autos, looking at the Honda Accord and Camry and all that. That’s totally different. I have a very different answer, now, about the value of that mail impression. It’s funny that we actually cut our teeth on monetizing social inventory, but it’s funny that everything that used to be true about Yahoo! Mail, in terms of its massive leadership position in display, has gone to Facebook of course.
Anyway, getting back to the list of three. There’s data, social inventory, and social objectives – meaning the goal is to get someone to “like” a brand or to Tweet about the brand. We’ve leveraged all of these, and sometimes, advertisers tend to conflate them. When they think of social, they think of all three of these at once, but we run campaigns with a social objective, like getting someone to “like” a brand on Facebook, but we run traditional banner display ads to make that happen.
What I would say about Facebook or any space you’re going to buy, is, “Never mind what you thought would work well. Let’s just see what does work well.” Then, on the social data front, I would say I don’t have stories where social data is a game‑changer for a campaign. There are certain companies that focus only on social data, and they’ll tell you once you have social data, you can forget demographics, behavior and context. All you care about is the social web of connections.
As far as I can tell, I don’t think that’s true. It stands to reason that there would be certain campaigns where the product itself may be more social in nature, where if you have somebody who’s become a user of that product, then their friends instantly become good prospects, but those are relatively rare.
“What are the odds?” You might answer that, “Well, we’re all graduating from college and we’re all buying cars,” but then I’ve got you at demographics. An advertiser can imagine the demographics that are leading to that buy. It’s not just the fact that you’re friends – that’s our view on social. For us, we haven’t been trying to run social‑only campaigns. We’ve been trying to blend social, video and display into single campaigns where we can then have flexibility to do our own bottom‑up media optimization or media mix modeling. Essentially, if we just buy the one next impression that looks the sharpest from a performance and cost perspective, then you essentially get – bottom‑up – the optimal media mix.
It’s simple. I want to keep customers happy. In phase one of the company, there was this rabid focus on making all the campaigns successful and building a reputation. In Phase Two, we’re beginning to scale the company pretty significantly now. We just got 160 employees, and we’re going to pass the DSP companies, if we haven’t already, in size pretty soon. We intend to double our employees in the next year. We’ve been growing faster than them recently.
We have 11 offices now. As CEO, my own focus is making sure that not only are we the best media partner our customers have worked for, but we’re the best place people have worked.
Beyond that, we’re expanding into Europe. There’s a vertical focus in the US. A strategic effort for us is simplifying the story about what we’ve done and getting it out more. In Asia, we’ve got some significant discussions underway about entering that market. We think there are segments of the Asian markets ‑‑ including Japan ‑‑ that would be very amenable to the Rocket Fuel approach.
By John Ebbert