The Alchemy of Search

Search Engines figured out a decade ago that allowing a single company to dominate the results was a bad user experience. Even if a company was algorithmically determined to have 100 relevant page results, the engines limited their appearance in SERP to one or two listings. Google improved this practice by indenting the second result for the same site and displaying it below the first, making it easier for the user to navigate directly to the page they want.

UI changes by Google over the last few months have seemingly lost track of this simple concept and created an embarrassment of riches for some searches. Specifically, Google’s rules appear not to consider the interaction between Local, Sitebox and regular results.

Consider these searches for Berkeley Toyota and Apple.

Berkeley Toyota is clearly the BEST result for the first query; they shouldn’t get three listings in local, four more from a Sitebox and the top natural site below that. Depending on how you count, that’s the top five spots and eight listings.

Apple gets six unique listings and ten links before the first non-Apple listing appears. Below the Wiki listings, Apple sub-domains scores another three of the next six listings and two more go to Mac.com. That means Apple owns 11 of the first 14 listings.

So, Google, how about fixing your rules here?

1. If a site is given the SiteBox/Onebox position, that should be the only result on the page from that domain. At minimum, remove the natural search result.
2. If a site is showing in local, it should suppress the OneBox listing. There is no point in showing both.
3. Even if a company shows multiple locations in the same city, only use one spot in the local search.
4. Sub-domains are considered “separate” sites, but if each sub-domains gets two spots in the SERP, even the purest White Hat SEO is going to be tempted to create five high quality “sites” and own the top ten.

By now we hope you’ve heard of Enquiro’s Eye Tracking Report and have been inspired as much as we have by the results. Interestingly, in this month’s issue of Business 2.0, a related study aimed towards gauging the effectiveness of TV ads by measuring the saccadic motion of an ad viewer’s eyes, uses a technology that we think could potentially be put to similarly good use on web-based content:

“The system, called eMotion, uses a PC-connected camera to measure the saccadic motion of viewers’ eyes—the subtle eyeball vibrations that increase when we see something of interest. PreTesting follows up with an interview to determine how much the viewers remember. Then a proprietary algorithm gives ad execs instant feedback on how well the spot worked.”

“Viewers should be engaged but not so visually stimulated for the full 30 seconds that they don’t take in any verbal information. ‘This could be revolutionary,’ he says. ‘Advertisers will finally be able to understand which ads are working.’ Weinblatt’s favorite example, from his beta-testing, was an ad for Dr. Scholl’s massaging gel insoles. It featured a leggy model emerging from a swimming pool, which sent the eyes of his male subjects vibrating. When the model turned out to be wearing high-heel shoes, his female subjects got excited. Then the ad cut to nothing but text and voice-over, letting the information sink in.”

As far as integrating such technology within the web environment, it could potentially fill the gap between “where the eyes go” and “what the mind thinks” when they go there. So when extending eye tracking beyond just identifying hotspots to examining actual levels of interest in the areas that the eyes are focusing on, we get even one step deeper into the mind of the visitor. And for video content, flash tutorials, or multimedia ads, the implications are very exciting.

According to the article, such testing is available for $2000 per TV ad – and with production budgets averaging $381,000 per 30 minute national TV slots, it is quite a small price to pay for effectiveness measurement in the context of television advertising. Obviously, for web applications, a web-specific algorithm would need to be developed (if one hasn’t been already), and a lower price point would make it a bit more feasible for the web industry. Look out heat maps, here come thought maps…

So we finally see PPA testing begin, and Google is first out of the gate. Before everyone jumps for joy about the promise of a (potentially) click fraud-free world, let’s consider for a moment what is actually occurring within the Pay Per Action model and the inevitable questions the model raises, in the context of converting directly from the Pay Per Click environment.

First, let’s see how Google describes the test pricing:

“You define a fixed amount that you’d like to pay for a completed action based on the value of that action to your business. You’ll only pay when that action is completed, not for a click or impression. For example, you may wish to pay $1 every time a user fills out a lead form on your site and $5 when a purchase is made.”

With PPC pricing, the advertiser does control the desired per-click amount based on the value of the click to their business (detailed ROI metrics are assumed for this discussion), and the desired spend. The advertiser ideally has a basic conversion metric that sets these budgets and so has an idea of what each “completed action” – a form submission, or a sale – costs them in PPC. The advertiser also has the estimated figure for what they can spend on a completed action, and working with these figures is the art of PPC management.

In their current PPA test, Google is telling the advertiser that the advertiser can use this metric, which we refer to as the advertiser’s Cost Per Acquisition (CPA), to set their “action” price, instead of bid price. Since it is a safe bet to assume that many if not most advertisers spend more than $1 to $5 of PPC advertising on Google to generate a single completed action, we can see that this test is being done at a discount* to test the platform, the implementation, and to see how well the traffic in their network is converting into specific kinds of “actions”. (*Disclaimer: testing on the Content Network arguably creates a lower traffic value, so the perception of the discounted testing is somewhat debatable).

Why is this meaningful?

First, let’s take a look at the data Google will be obtaining from its network of Content Partners. Publishers in Google’s network will receive, it appears, flat rate compensation. In exchange, Google will receive all sorts of conversion information about that publisher’s web site – as well as demographic information about that site’s users based on the actions of visitors clicking out to the PPA ads. While on one hand this sounds promising for the optimization of the ad network, will Google freely exchange this data with its publishers? Food for thought.

Next is the issue advertisers must address if seriously considering moving towards the PPA model – and that is whether or not there can be a comfort zone within disclosing to Google, or any publisher, for that matter, exactly how much an action is worth to an advertiser’s business. And if that action is a sale, then the issue is whether there can be a comfort zone within disclosure of revenue from same.

In theory, it is possible to envision PPA driving the cost of Google advertising up even higher for advertisers than it is now, because although the competition between advertisers will exist much the same as it does in PPC, once Google knows the actual advertiser-specific and industry-specific margins there then exists the distant, but yet real possibility of minimum bid-setting based on this data. Of course Google has done nothing of this kind to date, but there’s nothing like good hard evidence to sweeten the pot .

In the end, will you be converted? Will we? Let’s wait and see.

True, search is about conversion. But not just in the typical sense of converting searchers into visitors, and visitors into buyers. The search industry is also about conversions of many, many kinds: from one pricing model to another, for example. Conversion from black hat – to gray hat – to white hat. Conversion from: “click fraud doesn’t exist” to “yes, it does“. Conversion from: “SEO is Bull” to witnessing first-hand the benefits of good old white hat SEO. Conversion from what’s tried and true, to what’s awkward and new.

These are just a few of the conversions we deal with in search, and we’ll be digging deeper into these and all of the new conversion matters as the industry continues to grow, transform, and transcend itself. This is, after all, Alchemist Media.