What’s Better Than a Database of Intentions? Actual Intent. How Social Media Advertising Out-Targeted Search Engine Marketing

John Battelle described search engines as the “databases of intentions” – when a user enters a query into a search engine, that user is telling the search engine something about his wants and needs. Indeed, the entire point of a search engine algorithm is to decipher this user intent and serve up the most relevant results (and, of course, to stymie search engine optimizers at the same time).

You would think that a database of intentions would be the ultimate targeting opportunity for marketers, and – until recently – I would argue that this was indeed the case. The problem, however, with this database, is that it is an inferred database. Sure, we can infer that a user typing in “male pattern baldness” is a man who is balding, and in most cases we’ll probably be right. But what of the person who types in “mortgage” – is he (or she) looking to get mortgage quotes, learn about the current housing crisis, do academic research, get a job in the mortgage industry, pay their existing mortgage, or something else entirely? And then there are words like “laker” which could be about the Los Angeles Lakers, Lake Trout, or someone with the last name “Laker.”

As search marketers, we end up doing a lot of inferring. Often the difference between a successful and unsuccessful campaign rests on our ability to correct determine which keywords have the right inference for our campaigns. Since every search marketer is playing the same game, the result is that keywords with clear inference tend to receive much more advertising competition than keywords without. This is the difference between buying the keyword “Miley Cyrus” and “Buy Miley Cyrus CD.”

Wouldn’t it be great if users could tell us more about themselves? Their interests, their demographics, their personal history? To some degree, this information is available through behavioral targeting, or from user registration information on portals like Yahoo. But the behavioral targeting to date has only been broadly applied – you can choose a particular age range (18-35), or a particular geography, or sex, but none of this really presents an opportunity for personalized marketing.

Advertising on social media can, and likely will, provide the first opportunity for truly one-to-one marketing. The beauty of social media (at least right now) is that users have an incentive to provide lots of honest information about themselves. Think about your profile on FaceBook. You add your educational background, your interests, or relationship status, your current employer, and so on. You do this because you want to connect with old friends or meet new people like you. Indeed, the fact that your friends will see your profile is a further incentive to be honest since any wanton lies would be seen by people who know that you are lying.

This is much more valuable information than the behavioral targeting that can be gleaned from a user profile on Yahoo or from user-entered registration data on non-social networks. It’s commonly accepted that users who are forced to fill in personal data purposely lie about their demographics. As one columnist put it: “Users lie to protect their privacy, they lie to protect their identity, they lie because they think their data will be misused or shared with third parties, or they lie because opt-in/out policies are misleading or mistrusted.”

But at least for now, most social media users don’t lie. The result is a goldmine of not just inferences of user intent, but user-defined extensive descriptions of their intent. And this presents the first opportunity for marketers to truly create micro-behaviorally targeted campaigns.

Here’s an example of the micro-targeting currently available on FaceBook. Let’s say I want to sell Iowa Hawkeye football tickets to football fans in Iowa. And for whatever reason, let’s say that I want to upsell these users on a dating site. Check out the targeting FaceBook offers for this unique and granular user set:
80 people are in my target group – now that’s a narrowly tailored audience. Granted, this targeting excludes people on FaceBook who have not fully filled out their profiles, but you can only assume that over time the percentage of people who complete their profiles will only increase.

Much of this targeting can be done on Google – you can buy the exact match “Iowa Hawkeye Football Tickets” and geo-target your ad to Iowa only. But the ability to serve this ad only to Iowa graduates who are men, etc, etc is not – and likely will not – be available in Google for some time to come. Indeed, I recently pointed out just how bad the demographic data currently available in AdWords really is in a recent post.

Of course, the flip side to this entire argument is that we are currently at a point in the development of social media where users still trust the social media networks and advertisers haven’t truly embraced advertising on this medium. As a result, users are still honest about their personal data. At some point – perhaps soon – the scales may start to tip; once users realize that their personal information is being used by advertisers, their honesty may diminish or – as has already happened several times on FaceBook – they may revolt against the use of their personal data.

I believe, however, that despite some very public missteps in the user of personal data, the social media companies still have the chance to use this data in a way that satisfies privacy advocates but also provides the most targeted advertising available anywhere online. Who needs a database of inferred intentions anyway?

 

Google’s Scary Targeting; Yahoo’s Scary Lack of Targeting

I had polar opposite targeting experiences today on Yahoo and Google.

On Google, I was using GMail to email my friend Kevin about setting up a lunch. The email chain was pretty boring stuff: are you free on this date? where do you want to meet? and so forth.

In the past, Kevin and I have also emailed about college football – I’m a big Iowa Hawkeyes fan, he’s a big UCLA fan.

For this reason, I was pretty impressed to see AdSense ads showing up alongside our banal lunch email chain with the following headlines:

• UCLA Fightsong Ringtone
• College UCLA
• UCLA Bruin Tickets
• Bruins Ringtones

Sophisticated targeting going on here. Somehow AdSense has associated my friend Kevin with UCLA (who knows, perhaps this is all he ever emails people about, but I doubt it) and therefore serves UCLA-related ads whenever I read any of our email chains.

In contrast, I regularly surf my “My Yahoo” page to check up on the various paid search blogs that I read, my stocks, my favorite sports teams, and the latest world news. Yahoo has a ton of information about me based on all these feeds and my emails. And yet, about 75% of the ads I get are the classic “free iPod” banners. Most recently, I was asked to vote on whether I liked Hillary Clinton (and ‘win a prize’ for participating), and to participate in a ‘test panel’ to receive a free dinner for two at Chili’s.

These ads are clearly “run of site” or “ROS” ads that Yahoo is selling for pennies per CPM. Not to toot my own horn here, but I’m sure that there are plenty of advertisers out there that would pay a lot of money to target ads to me. I’m a Silicon Valley marketing executive with a lot of disposable income (well, a lot more than the average American at any rate). You could target me with bid management software, CRM, conferences, travel, fine dining – there are many ads that might appeal to me that could pay Yahoo big bucks.

But instead, I get the product test panel ROS. Google’s remembering my old email conversations, and Yahoo can’t differentiate me from a 12 year old in Tulsa. And you wonder why Google’s stock is 20X Yahoo’s.

 

Google Launches Behavioral Targeting – Interesting, But . . .

Andy Beal posted on ThreadWatch this morning about Google’s plans to offer demographic targeting. As Andy notes, this comes “on the heels of MSN’s demographic-friendly AdCenter.” To quote Lee Corso, however, “not so fast, my friend.” Google’s announcement is interesting – and significant – but it is also quite a bit different from what MSN intends to do.

The interesting part is simply that Google, MSN, and Yahoo (in May or June) are all jumping on the behavioral targeting bandwagon. For a while it seemed like Revenue Science, Tacoda, and Kanoodle were the only ones who cared about SEM demographics and psychographics. Now that the big boys are joining the fray, you can expect to see search engine marketers get a lot more serious about their log files, Web analytics, and Nielsen or comScore subscriptions (Google, by the way, is using comScore data, which I have always thought was more accurate that Nielsen, so way to go Google).

But let’s be clear what this Google announcement is and isn’t. The demographic targeting Google is providing is for site-targeting only. Site-targeting is Google’s CPM product. With site-targeting, you type in a word (for example “Iowa hawkeyes”) and Google shows you sites on their distribution network that are relevant to that word. You can then opt-in to advertise on these sites, and you can bid your max CPM amount to show up.

Site-targeting doesn’t apply to Google, or to any search results, for that matter. It also excludes most of Google’s big distribution partners (like AOL), and as noted, it’s CPM only. So Google’s behavioral targeting really only applies to a limited segment of their network.

Also, this is really only quasi-behavioral targeting, in my opinion. The dream of behavioral targeting is that you will someday be able to choose your demographics and have your ad show up wherever someone with your chosen profile is surfing. I believe MSN once described this as such: MSN has information that a particular user is looking for a mortgage. When that user types in “basketball shoes” on MSN, you could still pay a premium and get your mortgage ad to show up in the ad results. Google’s site targeting is essentially giving advertisers access to a slice of comScore data about the user demographics of Web sites on the Google distribution network. This is cool and helpful – don’t get me wrong – but it is not true behavioral targeting, not by a longshot.

Still, kudos to Google for continuing to push the envelope here. I recognize that Rome wasn’t built in a day, you have to crawl before you walk, etc, etc. As John Battelle says, Google has an incredible “database of intentions” on user behavior. If at some point they decide to rely on that database (as opposed to the comScore database), and apply this to CPC ads, whoa, look out – that is some major behavioral targeting they can bring to the table!

 

Personal Information and Paid Search

Amidst all the concern about Google Desktop and Gmail compromising user privacy, it seems that folks have forgotten that Google’s competitors – in particular Yahoo, MSN, and Amazon – have far greater depth of user data, both in breadth and time period.

If you are a registered Yahoo user, you’ll start to notice that advertisements are eerily similar to Web sites you are visiting off Yahoo. For example, I recently bought VoIP service from SunRocket. As I struggled through trying to get the product to work, I visited the SunRocket Web site numerous times for installation instructions. These days, when I login to Yahoo Mail, guess what? Lots and lots of SunRocket banner ads.

And Amazon’s users are no doubt familiar with the “customers like you also bought” recommendations that permeate the site. MSN has fused behavioral targeting into the new MSN Ad Center by enabling advertisers to chose demographics to target to.

Generally speaking, there are three ways that these companies are getting this information: 1) When you register for Yahoo and MSN, you are asked for demographic information. 2) Cookies on your computer; 3) Collaborative filtering (matching your user behavior to that of other users and concluding that if both you and another user liked Book A, each of you will also like other books that the other has purchased).

Privacy concerns aside, this is a potential goldmine for search engine marketers. Right now, we are all “keyword focused.” We limit ourselves to keywords that are directly (or at least highly tangentially) related to the product we are selling. For example, if you are selling bottled water, you would no doubt purchase keywords like “bottled water” and “water home delivery”, you might purchase “soda alternative” and “healthy living”, and you would doubtfully purchase “iowa basketball” or “new light fixtures.”

But what if you knew that an MSN registered user had been surfing bottled water sites all day long? Even if that user typed in “Dr. Seuss” as his search query, you would probably nonetheless pay a premium to get your bottled water ad showing up on this user’s search results.

Again, if you put aside privacy fears for a moment, this is actually a win-win-win scenario for the advertiser, the search engine, and the consumer. From the consumer perspective, the ads that are shown are of the highest relevance. Creating an algorithm to serve ads based on actual user behavior, the behavior of similar consumers, and the actual search query entered into the search engine is far more powerful than simply relying on the keyword itself.

For the publisher, this sort of targeting is a way to increase CPC prices across the board. The CPC keyword market is highly inefficient, precisely because advertisers are left to guess which keywords are relevant to their target consumer. And whenever guessing is involved, it inevitably results in many advertisers missing the right keywords (hence the obsession with “the long tail of search”). By combining keyword buying with demographic and psychographic targeting, the search engines essentially can create an uber-advanced search matching system that guides advertisers to the right consumers, irregardless of that advertiser’s keyword generation competency.

Finally, for the advertiser, targeting eliminates the aforementioned guesswork of matching keywords to the right users. At the end of the day, savvy advertisers are concerned not with CPC cost but rather EPC (earnings per click). I would much rather pay $10 a click for a consumer who I know will buy $25 of product from me than $.25 a click for a consumer who will only by $.20 worth of goods.

At this point, you may be thinking “isn’t this already being done?” Well, yes, it’s true that there are behavioral targeting companies out there – Revenue Science, Tacoda, and Kanoodle seem to be the ones I hear about most frequently. It’s also true that MSN Ad Center offers limited behavioral targeting, though this is a far cry from the collaborative filtering of Amazon or the targeting of Yahoo CPM banners. So it’s true that behavioral targeting exists, it just doesn’t really exist for the majority of search spending – namely on Yahoo and Google.

And this is an area where Yahoo, MSN, and Amazon have a huge advantage over Google. Google simply doesn’t have this level of user data. They lack the registration data because they don’t ask for demographic/psychographic information when you sign up for their services; they lack the collaborative filtering because they don’t sell millions of products like Amazon; and they lack the cooking because privacy advocates would go crazy if they attempted to cookie all users.

In John Battelle’s book, The Search, he describes search as “the database of intentions.” In other words, Google and Yahoo have tons of user-created data (i.e., search queries) that indicate user intent. I would argue that search is but one ‘database of intention.’ Shopping sites like Amazon and Shopping.com have databases, portals like Yahoo have another, and search engines like Google have yet another.

As Yahoo, Google, MSN and Amazon struggle for paid search dominance (or, I guess according to Yahoo’s CFO, second place to Google), it amazes me that this hasn’t become core to any of Google’s competitors’ offerings. In Al Ries and Jack Trout’s great book, The 22 Immutable Laws of Marketing, they write that if you can’t be number one in a category, create a category that you can be number one in.

Google is the biggest search engine, that’s beyond dispute. And I doubt any of Google’s competitors will unleash a better search algorithm than Page Rank anytime soon. Amazon, MSN, and Yahoo, however, all have the ‘database of intentions’ to be the number one behavioral targeting marketing company. And because these are proprietary databases, no Google search algorithm can penetrate this data. In other words, it’s a defensible advantage, powerful, and unique.

Hmm. Maybe it’s time Google wannabes stop trying to build the best algorithm and start leveraging what they’ve already got!