Yaron Galai is CEO of Outbrain, a content discovery technology company.
Click below or scroll for more:
- Outbrain and Content Marketing
- The Publisher's RSS Feed
- Problem That Outbrain Solves
- Use Case
- On Links And Engagement Metrics
- Target Market and Market Size
- The Future
YG: Well, think about search. You have the search engine, which is the service you provide and that creates the "real estate," and the business model around - search engine marketing. The way we think about our platform is as a content discovery engine, and the business model behind it is content marketing.
As far as momentum, we feel there's definitely momentum around us. We'll see how big it can grow. In its current form, it didn't exist a couple of years ago. Marketed content has existed forever in different forms: on TV, special inserts in magazines and things like that - but nothing online.
Outbrain started in trailing RSS, and I personally was using Bloglines and thought, "That's the place where it should live." RSS as a thing has been degraded to plumbing, and it's not a thing of its own. I've stopped using any RSS reader. Content sort of discovers me through Twitter and being on publisher sites through our service. I don't think RSS matters much anymore for content discovery. For us, the principle when we started with was "Let's have our links show up where people are already consuming content." We know we didn't want to be a destination site or a Digg. We wanted to be where people consume the content. I thought it was all going to go RSS, but now I think that's not happening.
Well, it depends on what's in the RSS feed because RSS is the “Twilight Zone” of content publishing. It probably boils down to the question of whether it's full content or not. As a user, I love the fact that there are RSS feeds with full content - as a business, that can only work if RSS is not successful because if RSS is successful, that basically means, en masse, content is consumed without the advertisements to support it.
A good example [of the conundrum] - I love InstaPaper as a user. I get a lot of content through InstaPaper. I can't understand how that thing cannot grow. So, it's one of those only services that I love, but I try not to endorse it much, because I know as soon as too many people use it, it probably has to go away. It undermines the functioning model.
Do you feel the same way about FlipBoard?
I actually had a blog post on it. I think FlipBoard is in a “Twilight Zone.” I'm not a big user, but I can see the big value of it. It's pretty great. Long-term, as it scales, it disintermediates the publishers from the two constituencies that matter for them – one is their advertisers. They've got their content stripped, and probably get FlipBoard advertising. The other constituency is the reader, it disintermediates the publisher from the reader. Long-term, the good publishers won't want to play there, because they know that they're basically building their number one competitor - for both ad dollars and for reader relationships. It doesn't make a lot of sense. And, a lot of publishers that play with them [today] are probably part of the general tablet and FlipBoard hype. Still, I'm not sure how this plays out long-term. It's going to be interesting.
I think of Quigo (Galai co-founded Quigo which was eventually sold to Aol in 2007) as publisher focused. Outbrain is publisher-focused, I think. Why do you keep trying to solve problems for the publisher?
I'm seeing this with some entrepreneur friends that are in the second or third wave of their [businesses], and people keep going back to solving the same problem, but better. A great example is Buzzfeed. In my mind, they're the Huffington Post guys. They're basically doing a better Huffington Post. Maybe it’s just a tendency, but for me personally, I love reading stuff. I loved magazines. I'd say most of my higher education came from magazines rather than college. I never graduated, and what was simple in the old magazine days, both from advertising and consumption perspectives, was you had one editor that made good choices for you, and you were done. That all changed with the Internet. Now, we have a ton of great content, but it’s almost impossible to find with all of the junk content that's being produced. The signal-to-noise ratio is terrible.
The advertising around it is terrible, too. It's difficult to stay in the good content. With Outbrain, what I'm trying to do is basically solve this problem for myself, by developing this big system that looks at all of the data and finds interesting content for me, and does it in a sustainable fashion. This is my lazy way of solving the "What should I read next?" question.
So, the typical venture capital (VC) question for you - "What problem is Outbrain solving?"
I'm not sure about all start‑ups, but the M.O. should be to find a clear problem and solve that. It was a problem for me to find the good content, and separate that from all of the content that's not interesting or relevant to me. I'm not sure it's a problem the way anti‑virus [software] solves a problem. It is easy to put a VC slide or deck together where you have a slide for a problem, a slide for a solution and the business model. Here, it's more about how I'd like to see publishing being done or serving the reader, and just having good faith in that. If we do that, if we are the trusted next link for many readers, there is a good, healthy business around it.
In some ways, the concept seems simple. So, here's another VC question on “barriers to entry.” What is it that you guys are going to be able to do differently than your competitive set? Is it about getting scale right now versus other market players?
Well, scale is always difficult. It's definitely a barrier for smaller companies. Obviously, there are bigger companies in this space, where scale is not an issue for them. The biggest barrier to entry is probably the data going into the algorithms. Developing algorithms is a barrier, but smart engineers and mathematicians can develop algorithms.
I could hand you the printout of all of the Google algorithms, and you could launch a search, which would be nowhere close to where Google is. The difference would be the best algorithm, if it's not fed a lot of data; it is going to be pretty stupid. The algorithm is just as good as the data fed into it. That would be one big barrier to entry, because that you just can't code.
Another big barrier to entry on the paid link side is just getting the inventory of buyers that are interested in moving traffic to their sites. The quality of an algorithm is as good as the data that's in it. In a similar way, the quality of the links are just as good as the index that's behind it. Those things are trickier to kick-start. There are some barriers there.
So, definitely the side we start on is the publisher side. The process is simple. The publisher installs our service, and we manage or serve the readers the recommendations for what's the next link we think would be interesting for them. So, we started with that - no paid links, no revenue model, although we knew what the revenue model was going to be. That's the distribution side. That piece of the service is free, and publishers that install it get an immediate page view lift on their site, because we recommend a lot of their stories.
Typically, publishers that install our service are going to see up to about a 10 percent page view lift on their sites for free. That's a simple beginning. That favorably compares to many of the related link services, which typically get a 2 percent click per rate. We don't think about our service, by the way, as a related link service. We think about our service as an “interesting link” service. Interesting links are sometimes related, and in many cases, are not related at all. They need to be interesting.
So, that's the first part of the business. Then, with publishers that choose to have links that are outbound and are pointing to content elsewhere, we'll turn on those links that point to content on other sites. We get paid from those other sites on a cost per click basis, and we share the revenue with the publisher that we're hosted on. It’s pretty simple on both sides.
Are there cookies involved?
There are cookies involved. There is a lot of personalization that goes on with the links that we show each individual reader, based on their past behavior and the content that they consumed. For example, the stories that you've read, whether through our service or elsewhere, we're not going to recommend those to you. That's all cookie-based.
We have decided to silo the use of the cookie data for each site individually. You might be on the USA Today sport section each and every day, and on USA Today, you'll get more sports recommendations because of that. But if you go over to Fox News, you're a different kind of person, or a different cookie for us, because we siloed the data per publisher.
Got it. Why do you do that? Why not do it on a co‑operative basis?
Long-term, we will. Our guiding light is what's most useful for the reader. I think that having the most comprehensive data set that we can is going to serve the readers better. We're going to be able to better tailor the links for each one. So long-term, that's absolutely the way it should happen. We're in an era of a lot of misperception on cookies and data being stolen by rogue ad networks, stuff like that. Short-term, we just wanted to absolutely err on the side of publishers' concerns on this, and not even have that discussion for now. We'll definitely, when the world calms down about the cookie craziness, try to entertain the idea of making this better for everyone. For now, we've decided not to.
Maybe that answers my next question, which is - where do you think Outbrain fits in the world of audience buying?
We get a phone call a week from folks that are in the audience and data business. We have a unique footprint on the web at this point, and our data is valuable. But, we're absolutely not in that business: we don't sell or buy data. We don't sell audiences. It's the reverse Catch‑22, where one of the reasons we get all these wonderful publishers to work with us in this great distribution network is that they trust us not to do any of that. That trust is paramount for our business, and we're absolutely on the publishers' side and would not do anything to undermine that trust. I know there's probably a lot of revenue to be picked up and we don't care about it.
Just curious… In general - what do you think about retargeting?
As a user, I feel like I'm in a dark alley and I have a FBI helicopter with the lights on me and I can't run away because it keeps hunting me down. I feel like a hunted fugitive. As a business, it's probably a good business.
Our formats are always links. We're always a link to the next most interesting piece of content. Then, there are visual versions of it, so a link with a thumbnail.
Another frontier we're exploring these days is video. We historically have done just articles, but many of our customers told us, "You're the content recommendation engine for our content.” So, getting into that, and linking into the next video is a big area for us.
Then, the mobile tablet devices are an interesting area. That's not necessarily a link per se, it's sometimes swiping and things like that. We're innovating there as well. We did release a version of our service for mobile websites a few months ago. I love that it's a simple link. On small screens, publishers are having difficulty migrating anything from their websites. They can't put navigation, banners or ads. So typically, on their mobile sites, they put just the content itself.
But our service migrates perfectly to anyplace where you can place a link. So, some of our publishers are using this now on their mobile websites and the results are phenomenal.
Phenomenal from what perspective? From the revenue or from a driving traffic?
We can never get ourselves overly excited by revenue being phenomenal - what we're excited about is engagement. If our engagement with the readers, or our service improves the engagement, that's what we get excited by. On mobile sites, what we're seeing is the engagement level with our service is significantly higher than it is on the web - about 60 percent higher engagement on the mobile sites. It's fairly simple because on the [PC-formatted] websites you have a lot of stuff competing for the user's attention. On the mobile site, you have nothing competing for their attention. It's basically the title, content and then our link. Our links do get a much higher engagement rate there - that obviously affects their revenue piece as well. So, our monetization for customers on our mobile sites is something that's exciting for them, especially if they have trouble monetizing that at all. But for us, it's all about, "Did we put a more interesting link and get higher engagement?"
How do you measure engagement?
We measure it in two ways. The first and most obvious one is click‑through rate. Does the click‑through rate go up or down? What’s not obvious is how to develop algorithms that improve click‑through rates. What we found when focusing just on click‑through rates is that the algorithms tend to prefer the sensational titles – and, you're kind of on a slippery slope for sensational titles. The other important half of our algorithm is focused on engagement after the click. So, after we deliver, did the reader stay engaged on the content? For that, we look at “time spent” on the page, and how many more pages or articles they consumed. We factor all that back into the algorithm. For us, when an article gets a high click‑through rate, but people bounce from it, that’s going to be pulled out of the system. We intentionally reduce our click‑through rate so that we improve the metrics of post‑click.
I look at it from the bottom up, and I always struggle to answer top-down questions like market size. So the answer is - I don't know what the potential market size is, but I'm pretty confident that if we create lots of value for people consuming content, then there should be a nice potential market for us to play in.
I also don't know how to quantify the market potential based on existing market size, because most of our buyers typically haven't been spending on things like Outbrain anywhere else.
With that said - the market we are going after (or more precisely - trying to create) is that of content creators, which wish to get more exposure for their content. Historically, content owners are traditional publishers, and those are certainly the primary target market for us. But these days the lines are blurring a bit, and there are other types of content creators, including bloggers, marketers and brands. For Outbrain the target market is fairly simple then - if you're producing quality content, we'd like to provide you with the means for getting exposure with a quality audience, at scale.
We do. We live within the constraints of editorial tone. It's a constant struggle of finding what the right balance is. There are publishers that fully trust us to serve and let the algorithms go to the extreme of what their audiences find to be interesting. There are many publishers that want to stick to their editorial tone, which I totally respect. There are some places where they're disappointed with what their audience is actually engaging with. They think they know them better, and they don't want them engaging with the stuff that they don't feel is right for their audience. It's a constant struggle between trying to educate some folks in editorial, and explain that relevancy.
For example, it matters a lot in the keyword world where, if people type in their keyword, you have to be relevant. But, in the content discovery world, relevancy might actually be bad for finding an interesting link.
It's sometimes a difficult thing for editorial to understand. An example we give is if you're reading an article about, let’s say, “banks settling with the U.S. government” from the past day or two, the perfectly relevant link to put out now is going to be one of two options: first, the same story from another newspaper. That's a relevant one and, as a reader, as uninteresting as it gets.
Another relevant link, as far as contextual relevancy, would be a story from yesterday saying the Obama administration is doing nothing about the bank settlement, for example. The contextual relevancy is perfect, but I just read a story of newer news to the contrary. I'm not interested in the old news. Those are examples of what editorial sometimes thinks should be in there - because of habit and we're in a Google world. It's our job to educate them.
The other case is striking a balance between making sure that we stick within publishers’ editorial tones and there are not two different things going on in the page. It's about finding the balance, but we're definitely part of the editorial world.
Why don't you just start recommending to the New York Times what they should put on their home page? Could it work like that, eventually?
Yes and no. We don't think of our service as a news service. We used to think about it that way, but we don't anymore. It's a little bit nuanced because a lot of our distribution, a lot of our “real estate,” is on news pages. But, we don't think about Outbrain as a news service. I believe that an algorithm is never going to be able to replace a human being in making a good judgment call on what's important and what's newsworthy. It’s always going to be done better by a human editor, and the publishers that work with us know how to do that better than we would in algorithms.
It's a fundamental problem that cannot be solved with algorithms. The amount of time and data we need to make a smart decision about whether something is worthy of recommendation or not usually makes most hard, breaking news obsolete. We try to focus much more on finding what's more evergreen or less breaking news, and make those recommendations.
In terms of milestones, looking a year or two out, what could you share in terms of things you'd like to see Outbrain accomplish?
Regarding sharing the future roadmap, we're not very good at it. We do one thing and we try to do it better. Since we first launched the service, we've more than 10 X‑ed our average click through rate on the network - from about 0.6 percent to over six percent now, across the network. That really excites us, and there's probably more huge growth ahead of us because we have a lot of ideas on how to make the recommendations better.
As far as needs, there are two areas where we always look to invest more, if we can. One is improving the network’s recommendations - and that's through better algorithms, infrastructure, servers and stuff like that. We have decent revenues at this point so that covers expenses nicely. The other area, which is an investment area for us is international.
Last year, we expanded into three countries in Europe: the UK, Germany and France. The system supports about 40 languages at this point, but supporting a language is not equal to being active in a market, working with publishers, working with buyers and all that. That's what we changed this year. Of those 40 languages we support, we put expanded offices in three markets, and plan on more. It usually takes a year or two until a market sustains itself in revenues. For that, it's good to have our investments (Outbrain raised $35 million late last year.).
We have also acquired a company in the past - Surphace from AOL. We have nothing specific planned, but when there are good opportunities, we'll definitely look at making more acquisitions so it was good to raise a war chest.
By John Ebbert