If you don't have a website of your own to monetize yet, don't sweat it! We've created a huge tutorial on how to create a blog in less than 20 minutes. If Taboola/Outbrain sound like they may be for you, click the button below and start your own blog!
During the past couple of years, websites who rely on display advertising to monetize their existing traffic base have had a number of additional opportunities made available to them. It’s no longer AdSense or bust; there are now several creative monetization solutions that are available to publishers both big and small. (See: 7 Examples of Creative Monetization in Action.)
One that has seen widespread adoption is the sponsored content model. Many of the largest sites in the world, including ESPN.com and CNN.com, now have sections of their pages dedicated to “Content from Around the Web” or “You May Also Like.” (Note that we’re using “sponsored content” to refer to the widgets offered by companies such as Taboola and Outbrain, and not native advertising.)
These sponsored content networks work in the same way that traditional ad networks such as AdSense do: they match publishers looking to monetize inventory with advertisers looking to get traffic to their site. But unlike many traditional ad networks, the buyers here (i.e., the advertisers) are often just other publishers looking to drive traffic to their sites. In other words, a lot of publishers are on both sides of the sponsored content revenue equation.
According to a Financial Times interview with Outbrain’s managing director in Europe, the network shares “about half” of any revenue generated with partner sites. The cost per click paid is said to range from $0.15 to $0.30 while click rates are in the neighborhood of 0.50% to 0.75%.
Using those (big) ranges, we can come up with some revenue estimates:
In other words, most publishers will see RPMs between $0.37 and $1.12 with Outbrain. That translates into $370 to $1,120 per million pageviews, meaning that you’ll need to have a pretty substantial audience in order to generate meaningful revenue from these widgets.
In order to partner with Outbrain, your site needs to have “over 10 million monthly US article page views.”
Performance data on Taboola is much harder to dig up; the company doesn’t share the revenue splits it offers, and details on performance are sparse. But we have a few data points:
That would translate into revenue per click (unique visits sent) of about $0.35, which is roughly in line with Outbrain (perhaps a bit higher). It also implies a click rate of about 0.05% per recommendation, which translates into 0.20% to 0.30% per page (since the Taboola widget often includes multiple recommendations). That gets us to about $1.00 in gross RPM (i.e., $1.00 in total revenue generated per thousand pageviews) before taking into account Taboola’s cut.
It seems likely that Taboola is sending more monthly unique visitors than the 24 million estimated by Compete, since not all advertisers create a custom tracking URL or redirect traffic through Taboola.com. In that case, the click rate would be a bit higher and the average CPC a bit lower–meaning that the Taboola metrics are likely in the same ballpark as the more concrete figures available for Outbrain.
There’s little available on the Taboola revenue share model, though it’s apparently better than what Outbrain offers.
If you’re thinking about carving out a section of your site to experiment with sponsored content, it may be helpful to know how this whole thing works. It’s generally similar to networks such as AdSense, in that revenue is earned (or advertisers are charged) whenever a visitor clicks on a piece of sponsored content.
The basic formula for calculating the revenue generated would be:
Visitors to your site x % of Visitors who see Sponsored Content x Click Rate on Sponsored Content x (Average CPC – Network Commission)
So how might one sponsored content network be better than another? They can’t control the number of visitors to your site or the percentage of those visitors who see their widget. (That depends on where you position it on the page.) What they can control–at least partially–is the click rate on the ads served. This can be done by serving up the ads that are expected to generate the highest revenue per view.
It’s a pretty simple concept, and similar to most standard ad networks. Sponsored content partners can increase revenue by only showing the pieces of content that are relevant to your audience and presented in a compelling way. There are two components of a “compelling” presentation:
In reality, a surprisingly high percentage of the sponsored content stories shown feature sub-optimal headline and image combinations. Check out this one from WSJ.com:
That Ernst & Young link, which appears in the top slot, is just terrible. The image is a poorly cropped company logo and the headline is about as interesting as, well, an accounting company whitepaper. The third placement here is also pretty bad; the image is way too small to make out clearly. These ads probably get terrible click rates, dragging down the overall performance.
Here’s another from CNN.com that features some pretty bad ads:
The upper half of the Forbes logo isn’t exactly a compelling image, and the headline below is equally bad. And it’s pretty difficult to even tell what the story below is; the image is way too detailed to be used at such a small resolution.
Below is a sponsored content section on Huffington Post that’s pretty good overall–several images of human faces and other high quality pictures. Mixed in there, however, is an ad that probably won’t get many clicks:
We think that’s the outside of an office building, but it’s pretty hard to tell. Either way, it doesn’t exactly call out to be clicked.
We’re taking an odd pleasure in pointing out bad monetization in action, so we’ll highlight one more example of a sponsored content widget that isn’t working:
Now, it’s possible that each of the sub-par implementations we bashed above is associated with a very high cost-per-click bid, which would justify it being in the rotation even if the expected click rate is terrible. For example, assume that the statistics on the Ernst & Young ad we bashed and the Lifescript ad (one of the better implementations from above: close-up of an attractive blond woman) are:
In other words, an ad may perform poorly from a click rate perspective but do well from an overall revenue perspective if the buyer of that ad is willing to pay a higher rate.
The point of the examples and commentary above is to explain how some sponsored content networks may perform better than others. The better a network is at identifying image + headline combinations that perform well on your site (i.e., get a lot of clicks), the more money it will make for you and for itself. Networks that show ugly, poorly designed ads have room for improvement.
We should note that there are a lot of good explanations for showing terrible ads. The network may simply be trying to collect the initial performance data, or it may be running low on supply of ads from top performers. (I.e., all of the “good” ads have already hit their daily limit for spend allowed.)
For most sponsored content networks, the average CPC is going to be in the neighborhood of $0.25, meaning that the publisher is going to get about $0.15 for each click.
If you’re looking to monetize a site via sponsored content, it will be definitely be worthwhile to test out a few different partners. Beyond the depth of advertiser pool, your network partner has the opportunity to make a big difference in your earnings by developing a superior algorithm to highlight the best performing ads and basically bury those that have no appeal.
For smaller publishers, Taboola will make more sense as a sponsored content solution since they have a much lower pageview threshold. For larger publishers the best solution will be to try out both; they payouts offered are very close, so it’s possible that different networks will perform better in different niches. Be sure not to commit to an exclusive long-term deal; give yourself the flexibility to try out a few different models.
Last Updated on