Interview of YieldBuild's Jason Menayan

With online ad networks slowing in growth and some collapsing I thought it would be a great time to interview the team from YieldBuild, an ad optimization service, to see where they saw online ads and ad networks heading.

What online ad markets are dominated by companies other than Google?

With its acquisition of Doubleclick, Google definitely does dominate both the largest publishers and the long tail of smaller content sites, but the fact it's the dominant player doesn't necessarily mean that it's exerted a monopolistic presence. At the high end, Doubleclick competes with (Microsoft's) Atlas, Yahoo, and (AOL's) Platform-A, while the long tail is served by a wide range of smaller ad networks, like Chitika, AOL's Quigo, Tribal Fusion, and Blue Lithium, in addition to the likes of Yahoo and Microsoft, which recently released its contextual ad network, Pubcenter. And when you go beyond traditional online media, the market's still open for mobile, games, and video. And, of course, lead generation (CPA) is a fragmented market that's certainly not dominated by Google.

Steve Ballmer mentioned how advertisers + search volume build off each other to create a higher yield. Do you see online advertising becoming a natural monopoly market?

I don't, because although economies of scale have an important part to play in establishing the pecking order among firms, there are other ways in which an ad network can successfully compete.

Google certainly benefited from being an early, aggressive mover in the space, and it is clearly the dominant player, but there is still substantial opportunity that is being capitalized on by other networks. Smaller ad networks fighting Google with a more generalized approach can still offer lower pricing, because heavy bidding with Google and limited high-quality inventory means that Google can't necessarily always provide the best value proposition to advertisers, so they offload to other networks. But there will always be other ad networks that either nail a specific vertical market well enough to be an attractive option for both niche advertisers and publishers, and smaller ad networks will also continue to innovate, creating engagement and pricing (payment) models that work better for some advertisers and publishers.

Google as a network can't do all these things - the market is just too big and too complex. But there is a way for Google to be the dominating player in the market by owning the marketplace, by opening up its platform to other advertisers, as it has done. Google doesn't derive any direct benefit by doing so, but one predominant ad delivery platform creates the liquidity in the market that makes Google's economies of scale matter.

What are some of the more innovative things smaller ad networks have done to gain ground on Google?

As I said before, Google draws tremendous strength on the economies of scale it's developed. But there are a large number of firms that can do some things better - either creative, targeting, deeper integration with advertisers - that can give it a leg up on Big G, enough to carve itself a profitable niche (at least that's been the case until now). Some have taken on the creative capabilities of traditional ad agencies, merged that with innovative, unique technology, and created online advertising formats that deliver better response/engagement. But targeting - being able to deliver a specific audience segment that advertisers want to reach - is something that smaller, vertical ad networks have been able to do better than Google or any of the larger, more general networks.

Do you see any ad networks playing the opposite role of Google? Google started a search engine to have an ad you think an ad network will ever build a search service around itself?

It's conceivable. I can see ad networks wanting to own properties that contribute valuable inventory. Search is an attractive piece of online ad revenues, but the competitive nature of search and the massive R&D budgets getting put into it make it unlikely that an ad network would be able to organically expand to include a consumer search service. It's much more likely that an ad network would build a search service through a partnership with Google, Yahoo or Microsoft, but it's a tremendous challenge to change people's search habits when Google does the job so well (although Ask, Yahoo, Cuil and others have certainly tried!). The closest I've seen is the one developed by Snap; their in-text links to pop-up windows include a small "Search the Web using" below the related content.

As Google builds more verticals (local search/maps, checkout, Google Product Search, book search, searchwiki, etc.) and adds features to their ad program (checkout logos, product links) do you see an eventual advertiser backlash happening against them?

Yes, although this is just a natural progression from partner to competitor as Google expands its feature catalog. I can give you one example from our own experience. HubPages, a site that we started in 2006, began as a partner with Google, offering users of the site AdSense revenue for publishing unique content on our site (we were the first site that used the AdSense API to manage this). HubPages has become hugely successful, and is a terrific revenue maker. Two years later, Google launches Knol, which is similar in many ways to HubPages. Naturally, it remains to be seen if Knol will ever become as successful as HubPages, but it's not surprising to us that Google would see how lucrative the business is and try to enter the market.

How many ad networks are typically in strong rotation on each YieldBuild client site?

It is hard to say, because there is no typical. We have a lot of publishers who just use YieldBuild to optimize their AdSense. Others already have a relationship with a display network like or Tribal Fusion, and they'll add that to the networks we optimize for them. We do typically recommend that publishers optimize one ad network for each two impressions a visitor is served from an ad network every day; this can be the case if a site gets lot of repeat traffic.

How often do you guys rotate through services to test them? Do you use earnings data cross sites to help improve yield?

The entire process is done through an algorithmic approach that uses performance data from the ads tested to determine the networks, formats, and layouts that generate the maximum revenue. YieldBuild is constantly testing, looking at changes, and adapting its algorithm to produce better results for our publishers.

We don't have any practical use across sites that can help any one site in particular, but we do monitor trends and can come up with more generalized trends like those here:

Some ad networks build added services in them to personalize the advertising experience. Do you see such personalization algorithms boosting yield?

I don't know of any data confirming it generally, but I can easily imagine that services which tailor each ad's creative or message to the visitor would boost yield. It's certainly been the case that our testing on HubPages with personalized/targeted campaigns generally do well, although the results are uneven.

What do you guys typically see performing better: text ads or image ads?

It completely depends on the site, page, and, most specifically, zone on the page. Sometimes a display/image ad on a page will do well, sometimes a text ad will, and often both will work well in rotation with each other. There is no way to know for sure unless you test; each site monetizes differently. It's certainly true that high CPM display ads are the holy grail, but there aren't enough of them (especially these days) to go around, so the goal should be to optimize your inventory with the best-performing ads, text and/or image, that are available.

How has the ad slowdown affected the network rotation ratios? Were smaller networks hit more than bigger ad networks?

I think it's too early to tell, but from our purview, rates are down across all networks. This is a pretty rough time regardless, though, since Q1 is weak generally. As the year wears on, I do think the biggest difference will be display vs text, mostly because text's generally CPC pricing model fits tighter marketing budgets better than display's CPM model, but we don't have the data yet to tell.

What baseline optimization ideas should a publisher implement before going to a third party for ad optimization help?

There are a lot of things that a publisher can do; some are simply applying best practices (like blending ads with the background, or embedding them in content), some involve a bit more work (testing). I've written a number of posts on our blog about how to optimize ads for blogs, forums and other sites. Naturally, a one-size-fits-all approach won't work best for everyone--you have to do more involved testing or use a service like YieldBuild--but it is better than blindly putting in ads in a haphazard manner and hoping for the best.

When does it make sense for a publisher to go with a third party ad optimization platform? Is the leading issue revenue, impressions, time, etc.?

I would say that unless optimization is a fun hobby for you and you enjoy it, or unless you're making little/no money and don't care about the revenue, then it's worth it to use a third-party optimization platform like ours. We haven't surveyed our users yet, but I'd guess the leading reason is to maximize revenue, while avoiding the hassle of tweaking ads all the time coming in second-place. Just finding the best ad sizes, positions and optimal number of ads to display for each page is very daunting to do manually, given its on-going nature and complexity of permutations. Beyond maximizing revenue and saving themselves time and trouble, platforms like YieldBuild also offer ad network management and deep analytics (comprehensive, consolidated reporting) which help publishers get insight into what inventory and traffic is earning them money.

Some ad networks (like Federated Media) often get quotes or other input from publishers and use it to help build the ad campaign. Can publishers work with those types of networks and YieldBuild at the same time?

YieldBuild doesn't do any campaign management; we're not a classic ad network. Rather, we support a number of ad networks that our publishers typically already have a relationship with. Federated Media is a bit of a different animal in that it works on an exclusivity basis; i.e. you have to agree to allow them to manage all of your site's ads, so I'm not surprised that they allow the publisher some input in shaping the campaign.

Blend vs contrast: which usually works best? When should a publisher consider using each.

The rule of thumb is to blend, especially above the fold and with white/light backgrounds. Below the fold, and with dark backgrounds, sometimes a color very close to the background works better, and sometimes a highly-contrasting, even bright, color works well. But often there's substantial benefit to nailing the exact right color, as in this example:

Google AdSense offers a heat map for ad placements. Do you think it is fairly accurate? What ad placements have you found that worked surprisingly well?

I would say it's a pretty good rule-of-thumb. It underscores that placement does matter, especially placing ads above the fold and juxtaposed/embedded in content. I hate to keep dropping links to our blog, but there is an example here that's interesting, because even at the handful-of-pixel level, the precise positioning of ads matters:

Do you feel that banner blindness will eventually carry over to other ad "units" to where advertising eventually has to leave the standard format size?

The IAB standard sizes have enormous value to the online ad industry because they help advertisers buy media at scale; too many custom ads just create too much friction for both advertisers and publishers, and relying on them would make the whole industry suffer. That said, although ad size is only one dimension that a viewer can become "blind" to (position, color, style, format all also matter, too), there will probably always be a market for custom ad solutions--there's an opportunity for combo packages that include a non-standard, custom ad product along with a lot of standard ads that publishers can slot in easily.

Excessive advertising on content can cut away at usability and site growth. What is the optimal number of ads/ad units that a publisher should display on a page? What are some ways people can include more ad units in their pages without making the pages look too ad heavy?

There was a study done on this recently that I blogged on, and if you're not using something like YieldBuild that makes that determination for you (YieldBuild will often serve less than the maximum number of ads per page, because this actually does improve page revenue), then I would probably do some sort of testing. Of course, it depends not only the number of ads, but their size, intrusiveness, how long your page is, etc. But you could always start conservatively, then slowly add more ad units and carefully monitor bounce rates. When bounce rates climb to an unacceptable level (minus ad clicks, naturally), then you could pare back the number of ads. This is assuming you don't have a way of A/B testing, which, of course, would probably offer better results.

Have you guys discovered great strategies for monetizing social media?

There is no one standard approach that works beautifully. Social media sites tend to monetize poorly, at least relative to their traffic - worse than original content sites. (This is something that even heavyweights like Facebook and YouTube are struggling with.) That doesn't mean that there aren't ways to improve what you are earning. Finding the right combination of ad networks, formats and layouts for your specific site can boost revenue. We have a large number of social media sites - some small, some very large - that are seeing impressive gains to their earnings by optimizing.

Who is the ideal client for YieldBuild? What types of publishers (site size, vertical, content type, etc.) can expect to see the biggest lift from working with you?

We actually work really well for just about every publisher. Naturally, larger sites will get through the training period faster, so they'll see improvement to their revenue more quickly. We've worked well for a lot of different types of verticals and content types: we optimize content sites, social networks, forums, blogs, and have seen success in all types. Occasionally we don't work well for a site, but we haven't determined any sort of pattern.


Thanks Jason. If you want to learn more about YieldBuild check out the below video or visit their site.

Published: January 15, 2009 by Aaron Wall in interviews


Add new comment

(If you're a human, don't change the following field)
Your first name.
(If you're a human, don't change the following field)
Your first name.
(If you're a human, don't change the following field)
Your first name.

New to the site? Join for Free and get over $300 of free SEO software.

Once you set up your free account you can comment on our blog, and you are eligible to receive our search engine success SEO newsletter.

Already have an account? Login to share your opinions.