While I am on holiday this week, through the magic of Internet time and custom tubes, my column isn’t. During this time off I’ll be taking time out for rest, reflection, and especially, a hiatus from one of my several jobs: Designated Undercompensated Google Ad Salesman.
What, you say? Google ads sell themselves? $14 billion in annual revenues would seem to suggest that yes, they do.
But recall that about half of those revenues come from non-search inventory: the contextual ads Google facilitates in partnership with tens of thousands of online publishers. Because—yes—the search ads by and large sell themselves, Google has put a ton of effort into moving the contextual inventory. And unbeknownst to skeptics, they’ve done much to improve it for advertisers. Although the financial breakdowns in annual reports are a bit difficult to interpret, a rough reading of the financials suggests that the share of overall revenues accounted for by the “Google Network” dropped again in fiscal 2007, to around 36%, from a high water mark four years ago near 50%. Google’s revenue growth overall has been driven primarily by the core: ads showing on Google Search and a few other Google-owned properties.
The first revenue growth hit Google took in this vein was largely based on smart pricing (an actuarial-type formula that paid low quality advertisers less per click) and stricter policies designed to weed the low-quality and nonconverting inventory out of the network. What was left gave a much higher-quality click to advertisers for any given click; even so, not all advertisers have been able to make the best of it.
Content-phobia is very real among Google advertisers. In spite of the many improvements in the platform and years of online advertising experience among clients, many a prospective client still comes to me with the casual comment that “we shut the network off – we found it just didn’t perform.”
Here I’ll look at good reasons for advertisers to have taken just this action, and then some reasons they should consider updating their thinking. Finally, I’ll offer some quick tips on how to get back on the contextual advertising horse.
Good reasons for content-phobia
The user interface deliberately tricks new advertisers. Industry consensus among experienced advertisers suggests that content targeting ought to be used selectively, and bid lower than search inventory. If asked directly, Google staff will tell you they now recommend advertisers create entirely separate campaigns for content—not even using the differential bidding option that would make it convenient to bid lower from the same campaign. But when you walk through a campaign setup you’ll find that you’ve been opted into the network at the same high bid as your search campaigns. You need to go back into campaign settings to either opt out, or opt into content bidding. Then, unnecessarily, the interface asks you if you want to set separate content bids “now or later.” Of course, you want to do it “now.”
Inappropriate publishers and content “spikes.” When you have content enabled in the conventional way, let’s say with a daily campaign budget of $200, sometimes you’ll suddenly start appearing on a high-volume, low-converting site like Bebo.co.uk, or MySpace.com. Not only is the wasted money annoying, it’s doubly detrimental to your performance because it gobbles up the budget from the side of the campaign you really wanted: search.
The “search network” also contains content. Think you’ve opted out of content targeting? Think again. If you have search “network” enabled in hopes of getting some search traffic from Comcast, AOL, and Ask, you may not realize that Google’s lumped in some other sources. Search-ish vertical sites like Kayak may send you traffic and you may want it, but they are really network sites, not search engines. That can be confusing. And Google tosses in AdSense for Domains partners here, without very good disclosure. This is doubly confusing again because in the content network proper Google now has excellent reporting in the form of a back-end report called Placement Performance Report, and opt-out capability for parked domains, adult content, social networks, and more. Not so for the muddy “search network.” (Expect this to change by year-end, a Google spokesman informs me. Further “iterations” are coming that will make the search network as transparent as the content network, I’m told.)
Control over placements has traditionally been nonexistent to poor through Google’s platform. And much of what we bid on behaves like remnant inventory that provides limited payouts to publishers. Google AdSense began as a clever hack that provided a short-term spike of revenue for Google and untold rogue publishers, and poor ROI for many advertisers. Due to this DNA, AdWords content targeting still doesn’t focus on premium inventory. Negotiating direct buys with publishers may be the only way to secure premium inventory and improved control over impact and timing, so as to best boost your brand.
Reasons to reconsider
Content targeting has improved since you last tried it. Leaving various display formats (banners, video, widget) aside, the conventional text ads are generally a better value today because of improvements in fraud protection, smart pricing, and network improvements. Of particular note is the fact that the more properties Google themselves own, the better they can watch the performance of content.
Reporting and opt-outs are vastly improved. You can exclude specific sites, various types of content such as parked domains, error pages, social networks, etc; you can also get conversion breakdowns if Google Conversion Tracker is enabled. Looking at the broken-out numbers may just prove to you that (say) parked domains convert about the same as the network in general. That’s certainly been my experience.
Classic content targeting isn’t the only game in town. Although specific site research efforts by individual advertisers tend not to be as efficient as they think, for those who don’t like to engage in Google’s mysterious “match on the fly” content targeting process, you can set up a Placement Targeting campaign that allows you to research and include (i.e., restrict your placements to) specific websites or pages. You can easily search by keyword, theme, or site. This function also allows you to create demographic filters where made available by the participating sites.
Expect further changes to the content program in general based on extensive beta testing currently underway at Google. The end result should be that Google will maintain a training-wheels version of content targeting for newbies, while offering a highly customizable dashboard for experienced media planners and data-driven marketers.
Dipping a toe back in the water
Use mirror campaigns to enable classic content targeting. Talk about back to the future! This was a tactic we were recommending five years ago! The safest way to get reacquainted with Google content targeting is to set up duplicate (“mirror”) campaigns in your account, keeping search shut off or bidding very low. Bid moderately to low on content targeting, keeping keyword lists of moderate length. In other words, you may be using a scaled-back version of your search campaigns, with fewer keywords per ad group, and not all of your ad groups. Track and monitor content separately, and consult back-end reporting in the AdWords reporting area every week or two to ensure that no weird sites are giving you traffic. Opt out of them (site exclusion at the campaign level) if they are.
Try placement targeting. Next, set up a couple of placement-targeted campaigns; consider bidding on a CPC as opposed to CPM basis (both options are now available). These may not perform as well as you think, because as a media planner, you’re one person; classic content targeting, by contrast, lets Google’s smart computers offer you a very wide reach (with less control).
Try brand experiments as an outlet for your contextual urges. Some things just don’t perform very well in search, but you’ve got excess budget and you want to boost your brand over time. Again, focus on how cheaply you can get the contextual inventory. Then, consider some banner testing. For text ads, consider putting your brand name right in the headline or body copy—tactics that don’t typically test out too well in the competitive search auction.
Try out different landing pages for content targeting. Many old-school TV-industrial-complex tactics don’t work very well from a paid search click; contests, high-concept verbiage, awareness campaigns, reinforcement of multichannel messages, and glitzy launches may tend to suffer from poor direct response in search. Again, if you have the budget available, go for wide reach in contextual ads at cheaper effective CPM’s, driving enough relevant traffic to these kinds of promotions to reach a tipping point in terms of buzz about your brand and your new initiatives. Here you can be less focused on immediate conversions and more so on long-term lift.
To my U.S. friends, may this “holiday season” be the best yet.
Opinions expressed in the article are those of the guest author and not necessarily Search Engine Land.