It goes something like this: While scrolling through the latest coverage on Syria, or doing research on a work project, you reach the end of an article, only to be greeted with the latest tabloid news about Kim Kardashian. With its teasing photo and promise of outrageousness, it takes all your willpower to resist. And you click anyway.
Over the past few years, content recommendation engines have exploded. Players include Outbrain, Taboola, AOL-owned Gravity and IAC-owned nRelate. Their widgets feature a mix of content from their own publishing network, links to other websites and links to advertorials or native advertising. (Read recent AdExchanger coverage of Outbrain and Taboola)
One reason these articles are so clickable is because they hide the fact that there is a transaction involved. Language like “We recommend” or “From around the web” obscures what’s going on. That skirting was enough to draw a ruling against Outbrain in the UK by the Advertising Standards Authority.
There’s a growing realization that in order for the field to be sustainable, something has to change.
AdExchanger talked to the founders of nRelate, CEO Neil Mody and Vice President of Product Oliver Wellington, about what they see as the problems in the space and how they’re trying to fix it.
AdExchanger: Your product started out as a way to recommend related articles to readers, but today it seems the space is optimizing for clicks, not the most relevant articles.
NEIL MODY: We and our competitors have done a really good job at figuring out what people would want to click on, but it’s like Kim Kardashian’s cleavage shot shows up in every article, because people tend to click on those things. Tabloid stuff, when placed everywhere, tends to get a lot of attention. We think that this is a problem in the space. It’s a short-term solution to get clicks vs. a long-term solution that is more about figuring out where the audience should go.
Could content-recommendation engines go the way of in-line text advertisements?
NM: The idea of having an underlined ad could have been useful, but somewhere down the line in that business, it turned into ads that made no sense. It got to a place where I didn’t even want to move my mouse around the page because I’d see some pop-up ad. We’re worried about the same things. This world has a lot of promise because it’s new and different, but it’s very easy for it to get spoiled or exaggerated or salacious, and that’s what’s happening now. The question is how can it become valuable long-term? We think it’s about getting back to what makes sense and what’s appropriate. That’s what we’ve been investing in for the past six months to a year.
How much control do publishers have over what shows up in their widgets?
NM: We built this thing called nControl that allows publishers to pick what should show up on the site, or on different sections of their site.
OLIVER WELLINGTON: There’s internal content, which they don’t pay for, and the partner links, where they make money. Sometimes the publisher might want to make more money and have fewer internal links. Other times, they might want more internal traffic, and all the links can be internal.
NM: We also have a white list, which can be 300 places where we’re OK with sponsoring links. They can also blacklist, and say these are a few hundred they don’t want.
How do publishers feel about the fact that people are clicking away from their content when they might have spent more time on their own site?
NM: There’s this paradox: Publishers say they want content that’s really similar to theirs. But then they look, and say, “Oh, those are all our competitors.” Then they x out all the content that would be relevant, and nothing makes sense. After that, they start saying OK, that’s fine.
How much do publishers make by adding in your widget?
NM: For many publishers, it’s the No. 2 driver of revenue below display ads.
OW: We’ve heard average display CPMs for publishers are in the $10-20 range. What we’re doing is anything from a $3 to $4 CPM depending on implementation and what we can run. At the same time it’s a growing space. CPCs have generally gone up in our industry, slowly but surely. When we started the business, getting CPCs of a nickel was good, but now we get CPCs well above 10 cents on average. We get CPCs of a dollar or more at times.
Tell me about the economics of brands putting their content out there vs. publishers.
OW: A nonbrand could pay 10 cents a click, and have a CTR of 1-5%. For a brand to get the same amount of impressions, they would have to pay 55 cents to a dollar a click. A publisher can make as much money linking to brands that have a lower CTR but a higher CPC. In that case, publishers are actually making the same amount of money for sending fewer people to those sites.
Can people pay more to get linked to on, say, CNN.com vs. other less prominent sites?
NM: We have to be careful about trying to rep sites. The risk there is the publisher wants to make sure no one’s circumventing their display ads on the site.
Tell me a little bit more about the kind of links that are giving the space a bad name.
NM: In the ad world, there are a lot of shortcuts. Publishers were sold on this promise of content. That it’s not a display ad – it’s engaging, it’s different. Then some of the ads that start showing up are not content in some cases, or point readers in another direction, all for motives not aligned with what the promise of the space could be. We’re different in that we sold our company so we’re really figuring out how to be a longer-term focused company, because I think the shorter-term focus, or ways to grab money are actually detrimental to the space as a whole. The main companies in this space are not acquired. AOL just bought a smaller player in the space, Gravity, six months ago. Outbrain, Tablooa – they’re not acquired, they’re still building their business.
Would you ever consider being a gatekeeper to the kind of content that can appear on your site as a way of fixing that problem? Outbrain has kicked out some content providers.
NM: We’ve thought about it, and we are not going to be the arbiters of what publishers put on their site. What we will do is give them the tools so they can figure it out. If you don’t want advertorial content on your site, you can pick and choose in our dashboard. We don’t get into the place of saying this should not be allowed. Except for noncontent. We’ve seen cases where the links don’t even go to content.
OW: Like a link to a mortgage calculator, or even an ad for a car. If you need to clean up your network every couple years, the process for content is broken. We’re building a platform for quality control.
How do you label sponsored content or native advertising in a widget that’s not on a separate site, but on, say, Parents.com?
NM: Our stance is once they get to the page on Parents.com, there should be a disclaimer.
What about the labels for the widgets themselves? A lot of them now say “We recommend” or don’t label the widget itself as sponsored.
NM: We’re following the IAB display model, where it should say “sponsored.” That’s actually getting cleaned up right now. We’re experimenting if that should be bigger.
What happens when you make the “sponsored” bigger?
NM: I can tell you without even looking at the A/B test that the bigger you make the disclaimer, the lower the click-through rate will be.
What kind of measurement do you offer, especially for native ads that may want to drive some kind of action?
NM: The widget has metrics on things like click-through rate. If they want, we can add in tracking for their software, like Omniture [Adobe Analytics] or Google Analytics. We enable that so they can figure out where the traffic is coming form, average time on site, uniques, all those metrics. We also let all our advertisers buy on whichever platform they want [tablet, mobile or desktop], or what geos they want. There’s been so much optimization on the display side, that hasn’t happened as much in our space, largely because it’s a newer space.
What kind of data you collect across your network?
NM: We have a stance that we don’t sell data to third parties, so all our data we keep in-house.
What kind of people come to you wanting access to that data?
NM: You’d be surprised. There’s probably not a week or month that goes by where someone doesn’t approach us for it. Typically who’s trying to buy that data are the optimization networks, like the display ad optimization networks, and/or search retargeting companies, or the cookie-aggregator companies that resell it again. They try to get everyone’s and package it up and be a one-stop shop. We don’t sell any of our cookie data. We let people clear the cookie, and only keep it around for 30 days.
What’s different about the space now vs. when nRelate first started? What’s the challenge moving forward?
NM: At the beginning, it was a different problem. A lot of the top-tier publishers and blogosphere products didn’t have great recommendations. Everyone was trying to be really contextual, but machines quickly figured out that if you don’t do contextual but do really salacious, click baity things, you’ll get more clicks. We’ve moved into a direction no one realized would happen, and the next hard problem is to solve it.
I think there’s a balance to be had between really contextually relevant stuff, that will help you get more information about an article you just read, and the noncontextual but interesting information. What needs to happen is balance, or even more so, appropriateness. There will be cases where there are entertainment articles, around salacious-type topics, and they should have those types of links. Then there’s a space where people are trying to read about the new phones coming out or shipping drones where it doesn’t make sense to cheat the reader into clicking over to something else. For us ad tech companies, it’s all about the value we’re providing.
Interview has been condensed and edited.
