Content recommendation doesn’t have the best reputation. It’s primarily known for spammy clickbait around the web.
But it doesn’t have to be that way, said Alon Rosenthal, CEO and founder of WhizzCo, a startup that allows multiple content recommendation vendors to compete for native inventory so publishers can maximize their yield.
Typically, publishers only work with one content recommendation vendor at a time on an exclusive basis, which limits their demand and the revenue they’re able to generate from native placements.
WhizzCo recently ran a three-month study to demonstrate the variance in ad performance across content recommendation providers by looking at 35 million impressions supplied by publishers working with the six top content recommendation solutions, including the market leaders.
WhizzCo saw a variance of 267% between the lowest eCPMs, which clocked in at $1.25, and the highest at $3.44. There was an even greater divergence for cost-per-click (eCPC) ads, which was just five cents at the low end and 44 cents at the high end.
“Publishers need more revenue and they need more transparency,” Rosenthal said, “because right now they’re leaving money on the table.”
WhizzCo works across 40 platforms and has direct relationships with the majority, including Google, MGID, LockerDome, EX.CO (formerly Playbuzz) and Revcontent who make their demand available both on a CPM and a rev share basis. Although WhizzCo doesn’t have a direct partnership with Taboola or Outbrain, it’s able to include their demand in its auction via a publisher’s integration on a CPM basis only.
Rosenthal spoke with AdExchanger.
AdExchanger: What’s the idea behind WhizzCo?
So, conceptually, like header bidding but for content rec?
It’s not a new concept, and yet our biggest challenge is to educate the market. Publishers do this for other forms of monetization, like display and video, but not for content recommendation. That said, the number of exclusive content recommendation relationships is starting to go down.
When the market began, content recommendation wasn’t a big part of the overall revenue pie for publishers, maybe 1% or 5% tops, not interesting enough for them to try and maximize. Also, no one had developed the technology to let content recommendation vendors compete for traffic. So publishers just implemented a single tag from a single vendor, usually either Taboola or Outbrain, both of which were trying to conquer the market by striking exclusive deals and offering guarantees.
What’s broken about the way content recommendation usually works today?
Any single vendor has a limited amount of ad demand, even the biggest ones. If someone visits a publisher’s site multiple times in one day and that publisher only works with one vendor, there’s a very good chance this person will see the same ads on each visit.
This is why for years people thought most content recommendation was spammy and low quality. We’re working very hard on technology to provide a better-quality experience, and the budgets we’re pursuing are going up each quarter. Advertisers are increasingly curious about it.
Still, that’s not the only reason people associate content recommendation with low-quality advertising. Most of the time, it’s the ad content itself not the repetition that’s a turnoff.
We’ve worked very hard on our algorithms to decide which topics are quality and which are not. If a topic is related to an unknown brand, for example, we might flag it and decide it doesn't qualify for our network.
There are two levels of quality we look at. The first is the vendor and the second layer is our own machine learning that goes through every article, topic and ad. For now, we don’t work directly with advertisers, just publishers and vendors. If it’s a vendor that we bring into an integration, it’s on us to ensure quality. We’re connected to enough demand sources and vendors that we’re able to pick the highest quality content from each.
What sort of targeting can you do?
We optimize for a number of different parameters, like geo, words on the page, the topic of the article someone is reading and the device and browser they’re reading it on.
What is the average eCPM and cost per click for publishers in your network?
CPCs can range anywhere from a couple of cents all the way up to 45 cents or 50 cents depending on the advertiser and how much they want the traffic. The network average CPC is 15 cents, although more premium pubs get around 20 cents to 25 cents.
With CPMs, it very much depends on the type of publisher, whether there’s organic traffic or a media buy, and how targeted the content is. A financial or sports publisher will have high CPMs because the audience is more targeted. On average, we’re seeing $5 CPMs, but it can go up to $6 or $7.
Even if you have direct partnerships with every content recommendation vendor in the world, there’s a gaping hole if you don’t have direct partnerships with Taboola and Outbrain. Have you been in contact with them?
I do see both of them joining us eventually. Although I can’t say which, we’ve had good conversations with one of them already. We even signed an agreement, but it fell apart at the last minute. In a year or two, I predict that the content recommendation market is going to look very different than it looks today.
This interview has been edited and condensed.