It’s no secret that social networks, like Facebook and MySpace, have had a relatively tough time selling their inventory. Much of the discussion has been around advertiser skittishness and poor targeting, but a new study by IDC suggests there is something different about social network users themselves: they are simply less likely, compared to overall Web users, to respond to online ads.

At first blush, it looks like it’s a simple matter of social networks enjoying a lot of repeat traffic, which dilutes CTRs and eCPMs. However, the study states that only 57% of users of social networks have clicked on any online ad over the previous year, compared to 79% of the Web at large. Furthermore, ad click-throughs led to far fewer purchases among social network users: 11%, compared to 23% among the Web at large. This indicates that it is not a matter of the social networks themselves or the advertisers that use them, since, presumably, social network users used non-social networks as well, and simply didn’t respond to ads on those sites as much, either.

That different types of online users respond to online ads differently is not new: I wrote last year of a “natural-born clicker” profile that clicks on ads far more than the general online populace.  I suppose it’s no surprise that if there’s a group that clicks more than average, there’s probably a group that clicks less often. Pity it’s a group that is so large and spends so much time online.

Why do social network users respond less? Here are a few guesses of mine:

• Because they spend so much time online on a particular network, ad blindness is more likely to set in. They are far more likely to know where ads are placed, and to filter them out.
• Since social network traffic is of a browsing nature rather than a searching/information gathering one, users are more likely to find ads irrelevant or uninteresting than those reading content sites
• Within the sheltered confines of their social network gathering place, ads, whether targeted or not, might feel intrusive of users’ sense of privacy

IDC concludes that social network advertising will continue to underperform until social networks take on a more content-oriented dimension like portals like MSN and Yahoo. It appears simply building a platform where users communicate and share with each other might not be enough.

YieldBuild optimizes online ad spots for publishers, maximizing revenue across networks, layouts and formats. But what if you’re an advertiser, and want to optimize the response rate to your banner ads?

There are a few companies that have created optimization services for advertisers that tweak design elements within the ad in order to improve click-through:

• SnapAds: Profiled by TechCrunch today, SnapAds takes creative from a network, agency or advertiser, and creates a large set of permutations with which to test. Backgrounds, calls to action and other aesthetic elements seem to be teased out of a multilayered source file provided to SnapAds, and are tracked for performance. The result is an ad (or small set of ads) that outperform the average. In one test, the SnapAds-identified ad format outperformed the original by almost 2000%.
• Tumri: Tumri’s AdPod parses out the core elements of a good banner ad: the product message, image, brand and call-to-action. Provided with a number of variants for each, AdPod can cycle through all the variations and determine those combinations that perform best.
• Personiva: Personiva’s AdaptiveAds, currently in beta, tout personalization down to the visitor level. Ads change depending on “environment, behavior and interest,” which I would take to mean context, although AdaptiveAds likely do something similar to Tumri and SnapAds: allow advertisers to test a wide variety of creatives with audience-targeted messages.

Why is this important? Banner ads are struggling to demonstrate their worth these days. And, while aesthetic concerns are important, they are not paramount, and leaving creative optimization to your creative team might not necessarily achieve optimal performance.

Before YieldBuild, I worked at QuinStreet, where the best practices team was constantly testing the effects of layout, format and other visual changes of ads, landing pages, and microsites. Sometimes subtle or seemingly inconsequential changes could have a significant impact on a piece of creative’s ability to engage a visitor. Thankfully, technology, such as YieldBuild for publishers, or Tumri, Personiva, and SnapAds, for advertisers, is making that work far more extensible and granular.

The premier tech blog, TechCrunch, just published an update on YieldBuild, covering the most important changes we’ve rolled out recently as we enter our public beta phase:

• we are accepting publishers of all sizes
• publishers can use our self-service wizard to provision YieldBuild code for themselves (you sign up via the application form, and we will send you a link to the wizard shortly)
• improvements to the algorithm have significantly shortened the training period (the time it takes for YieldBuild to yield significant lift to your revenue)
• we support more ad networks: in addition to Google AdSense, we now support Yahoo Publisher Network, BlueLithium, Advertising.com, ValueClick and TribalFusion, with more contextual and display network support coming soon

We thank Jason Kincaid at TechCrunch for the coverage and look forward to seeing more publishers enjoy great revenue improvements using YieldBuild, without the hassle or guesswork involved in trying to optimize your ads’ formats, layouts and ad networks manually.

2% growth from Q2 to Q3. Keep in mind that that’s a mix of contextual/search and display, and the former’s growth is more than certainly offsetting the latter’s drop. Assume flat growth for Q4, and we’re looking at a 2008 revenues total of $23.2 billion, a banner year and a 10% increase over 2007. But, if you’re inclined to pessimism, as is Henry Blodget, even that projection is too rosy.

Erick Schonfeld, covering online ad market trends over at TechCrunch, warns that the situation outside the U.S. is likely to be even more dire.

As the market for display advertising continues to look gloomy, it’s instructive to look at trends in the supply of search traffic now that demand (at least for search and contextual traffic) is up in the air.

I took a look at Hitwise’s most recent press release about search market share, and noted that Google gained search market share, at the expense of Yahoo, MSN and AOL, from September to October. Looking back over the entire past year, a trend is clearly developing: Google is slowly accreting search traffic and solidifying its dominance over Web search. And overall search engine traffic continues to inch upward.

What does this mean?  It means the relatively rosy predictions of healthy contextual ad growth aren’t off-base. The search advertising market includes both ads delivered directly on SERPs and those served on publisher sites via contextual ad networks (the largest of which are owned by the largest search engines themselves), and publisher sites continue to receive growing quantities of searches from the largest search engines. The chart to the right demonstrates that navigation via search engines (especially Google), despite consolidation of the blogosphere, substantial weakening of online retail, and competition from social networks, hasn’t slowed, and there is no reason why it should with a softening economy. (Searching is still free, right?) Combine that with the fact that performance-based, measurable-response advertising is the least likely to get cut from advertising budgets, and you have a relatively sanguine outlook…for contextual advertising, at least.

If you’re using a CPC ad network, like Google AdSense, on your site and are wondering where to position your ads in order to maximize click-through, here are some rules of thumb on ad placement that we’ve gleaned from our experience, and which was echoed in today’s Inside AdSense blog entry:

• Place your ads above the fold: Clickthrough is maximized when visitors see the ad(s) without having to scroll down.
• Juxtapose ad units: Place dissimilar ad units next to each other, so that they share a portion of a face with each other. For example, orient the short end of a horizontally-oriented leaderboard or banner against the top portion of a skyscraper.
• Embed ads within content: There’s a reason the popular medium rectangle (300×250) ad unit performs so well. It’s almost always embedded within the page content, so it tends to be within a reader’s line of sight.
• Place ads where visitors are likely to look and click: Although AdSense has rules against placing ads near images in a way that suggests there is a relationship between them, make sure you place ads next to navigational elements and other parts of the page where visitors are likely to look. Even better is to keep them near places where they will eventually click, so that both their eyes and pointer will be close in case the ad seems appealing.
• Make your ad units fit into your overall site design: Although it’s good to get visitors to notice your ads, you’ll naturally want to avoid having them interfere with the user’s experience. Obnoxiously intrusive ads will turn visitors off.

As the Inside AdSense team states themselves, every Website is different, so while following these suggestions should help you, only a rigorous optimization testing strategy, or YieldBuild, will determine the exact sort of positioning and placement tactics will maximize clickthrough and revenue for your site.

I’ve recently returned from the annual PubCon conference in Las Vegas, in which I attended the “Contextual Ad Program Vendor Roundtable.” Shuman Ghosemajumder, business product manager with Google’s AdSense program, described an interesting case study that underscores the adage that “less is more,” at least with respect to determining the right number of ads to run on a page.

Ghosemajumder recounted an AdSense publisher, a forum, that experimented with the frequency with which ads were running on his site, reducing them from running on every page impression, to one in every 10 impressions (I believe he initially did so to reduce their annoyance to forum members). He had expected his eCPM to drop, but, in fact, it increased. The conclusion drawn was that seeing ads on every page impression made forum members numb to them, but seeing them rarely made the ads stand out and command attention, thereby encouraging a click.

This study reminded me of a couple of conclusions that we had come to, drawing on our own data:

1.  several “good” layouts, running in rotation, generally outperform a solitary “great” layout
2. fewer ads on a page can outperform more ads on a page

.

Why the data bear these conclusions out is not exactly clear, although there does seem to be a benefit in keeping a page’s ad layout a bit unpredictable. Consistency is the bane of optimum ad performance when the bulk of your site’s visitors are repeats—they simply learn to ignore ads when they know exactly where they are.

When running YieldBuild, a publisher specifies the number of zones he/she would like to test, with the understanding that not all zones will necessarily be filled with ads. It is, in fact, often the case that zones remain empty when they could be occupied by ads, and the layout arrangement changes from page view to page view, even though YieldBuild’s goal is eCPM maximization.

As the Dow continues to hesitate about how low it wants to drop, and Congress muddles through the scope of the financial bailout program, those anxiously watching and forecasting trends in the online advertising market (and the large measure of the Web supported by it) are starting to sound the alarm bells.

Gawker’s Nick Denton, whose blog empire is wholly supported by ad revenue, has been leading the charge that online advertising faces a severe downturn through the coming year, his case explored in-depth in his doom-mongering blog post. In it, he presents his leading arguments:

• the online advertising market amplifies overall economic trends; it enthralls when the economy is humming along nicely, and bottoms out when the economy stalls. He refers to Mary Meeker’s regression that suggests a 3% drop in ad revenue for every 1% fall in overall economic growth.
• the US economic downturn will be as profound and slow to recover as other prominent national-level downturns like those experienced by Japan in the late 1980s and 1990s, and Indonesia more recently
• online advertising is, despite all claims to the contrary, not mature or measurable enough to avoid a downturn similar to that of Web 1.0

His prescription, in short, is to shuffle content and resources to capture verticals that are safer (technology over politics, for instance), squeeze more out of vendors and content suppliers, and unleash technology and creative resources to innovate.

Lost in this jeremiad is the long tail of publishers, most of which source their traffic through search, and who primarily monetize through contextual advertising and vertically-aligned niche ad networks. Denton, and the WSJ’s Peter Kafka, who posted data supporting the more dire predictions, hail from the world of big media, dominated by brand display advertisers. As I posted before, big (online) media has cause for worry. Publishers that target search traffic and align their advertising towards conversions are likely to continue to do well. After all, search traffic continues to grow, and Microsoft’s Silk Road API announcement only points to growing integration between the search giants and publishers who monetize through their contextual advertising services.

It remains to be seen how display advertising will attempt to forestall the damage inflicted by skittish advertisers, but I believe Denton’s final point about technology and creativity is his strongest argument yet.

Update: TechCrunch’s Erick Schonfeld takes a look at the top 4 online advertisers, sees low-single digit growth in Q3 and projects a flat market for Q4.

The bulk of non-contextual advertising, display advertising, is typically denominated as CPM, in contrast to the CPC of contextual advertising. Why? The bulk of display value’s advertising is to deliver a brand message to a visitor. Targeting of the visitor has primarily been a matter of matching a site traffic’s demographic and geographic information to relevant advertising; more sophisticated implementations have looked at a visitor’s browsing behavior on and off the site, to suss out their general interests and intents, with which to better target the advertising they’re being delivered.

But all of this targeting is predicated on the assumption that the visitor notices the ad. Maybe he scrolls past it in less than a second. Maybe it’s hidden behind another browser window above it. Maybe he shows a pattern of ignoring all image ads with cutesy animations in them. If that’s the case, the most targeted, $100+ CPM ad you’ve delivered to him has been effectively wasted. And in a tight economy, display advertising, most of all, needs to demonstrate value to the advertiser if it wants to avoid the chopping block.

Lotame (that’s loh-tuh-mee, a conjunction of locate, target, and message) is a startup that is seeking to address the issue of visitor engagement with display advertising, and has just landed a deal with WPP’s Mindshare unit to use its technology to better target ads for its clients. Why? According to a video prepared by Lotame, 20-30% of Internet browsing is performed on social networks, while this medium only captures 2% of the total online ad spend. A tremendous exposure opportunity, naturally, provided advertisers know they’re really getting the eyeballs they paid for.

How does it do this? It’s special sauce seems to be carefully monitoring an individual user’s behavior and screen environment. It claims it can notice when an ad is being obscured, and it can detect if a visitor’s mouse is moving or still. Such data gives the firm an idea of the time spent on user engagement with ads. The network can then choose to not display ads to visitors that are sure to ignore them or not even see them at all. It’s a clever technology, and technology is and will be a key determinant among those ad networks that survive the crunch.

The AdSense Optimization Team, which helps its publishers make their ads perform as well as possible, has released a blog post with the rule of thumb that ads that blend into your site template’s color palate tend to perform better than colors that stand out. This is exactly what we had suggested in our on AdSense formatting back in May, advice that we derived from our own testing.

The rules of thumb they suggest, which we corroborated 6 months ago, are:

• ads that blend into the page background tend to perform better (not always, but usually)
• borderless ads (or those with a light color) tend to perform better (again, not always, but often)
• try rotating the position and colors of your ads regularly (something that happens automatically if you’re using YieldBuild

We agree with these general suggestions, and would add that for ads below the fold, it might be advantageous to try a contrasting color instead of one that blends (although you won’t really know for sure until you test, or use YieldBuild to do the testing for you).