We're witnessing an explosion in the types of data that are available to improve marketing effectiveness. These new kinds of consumer information are intriguing because for years we have essentially been tinkering incessantly with the same behavioral targeting data points. Tiny improvements in performance make such a difference when your interaction standard is .08 percent, so all that experimenting was fruitful. But over time, the incremental performance lift declines when testing degenerates to dithering.
But lately, a host of exciting new data sets have been made available, both through the exchanges and networks. Here are five of these new data sets and how they might make a difference for you.
Values- and interest-based targeting
First up is values targeting, an area pioneered by Resonate Networks. The premise is that by understanding the values and interests of consumers, we can predict category and brand interest as well as future purchases.
The simplest use case would be in politics. (Bear with me -- I will get to consumer brands in just one sentence.) The fact that someone is pro-union is probably highly predictive of being an Obama versus a Pawlenty supporter. That's the basic premise -- and now forget about politics because there is enormous brand relevance to this methodology as well.
An interest in children's nutrition issues might make someone more likely to be a good prospect for Kashi cereals. Church-going might correlate with Chik-fil-A consumption. And beyond these gimmes, the people at Resonate say that there are a host of surprising correlations between brands and values. I think the richer your brand equity, the more likely that values targeting will make an impact for your business. Think about the first five values you associate with Coke. Now Pepsi. See how this can be relevant?
Resonate helps identify profiles with a high degree of likelihood to perform for your brand, and then purchases media directly from publishers to match your offering to these high-potential interest-based audiences.
Pop quiz: What's the most frequently used methodology for sharing content? The "like" button? Tweets? Nope. It's good old-fashioned cutting and pasting -- people highlighting a paragraph, copying the content, and pasting it somewhere.
According to Tynt, which has singlehandedly pioneered this new arena, we cut and paste far more frequently than we use any of the new-fangled tools and buttons. Data indicate that, on some sites, 6 percent of users cut and paste content during visits. And what we cut and paste says a lot about what we care about.
Tynt turns CTRL-C and CTRL-V activity into rich targeting information by semantically analyzing the content in a person's cuts and pastes using an advanced taxonomy. The company then combines what it gleans with more "traditional" types of data to create precise profiles of consumers and what they care about.
More than a million pubs quickly got on board with Tynt because its offering appends the URL of the original content when the consumer pastes the excerpt. This helps drive incremental referral traffic and SEO for the publisher. The company also offers an analytics suite that reports what users care about. This can drive greater page yield and can inform editorial departments of visitors' strongest content interests.
Here's what happens when I cut and paste a bit of the SF Chronicle's website (a Tynt partner) into an email:
When purchasing a house, most buyers have to finance the majority of the purchase price with a mortgage. The amount of money you put down upfront determines the size of the mortgage. Knowing how much of a down payment to save up for can be a tough decision.
Read more: http://www.sfgate.com/cgi-bin/article.cgi?f=/g/a/2011/05/17/investopedia52357.DTL#ixzz1NBugp9rL
The "read more" link is auto-magically added, making it more likely that a reader will visit the actual article. Meanwhile, Tynt semantic processing surmises that there is a greater probability that I am considering a home purchase or a refinance. That and other info collected through third parties helps banks, mortgage companies, and others identify me as a prime candidate for messaging.
Many of us use traditional retargeting -- the process of messaging to site visitors if they leave without making a purchase or taking the desired action. Retargeting campaigns tend to be very strong performers. After all, these people were interested enough in what you offer to visit -- but the limited amount of most sites' traffic limits scale.
Acerno (now part of Akamai Advertising Solutions) pioneered the concept of an online retailer data co-op, in which like retailers share their aggregated site visitation data with each other to identify more in-market customers. It's a means of growing the "retargeting" because everybody's site visitors are pooled together. The data co-op model actually comes from the direct mail business, where cataloguers shared customer contact information under the theory that a good L.L. Bean customer might also be a good Land's End customer. Cataloguers know that the possible lost sales against existing customers are more than outweighed by the savings of being able to mail to existing catalogue apparel shoppers versus cold prospecting. It seems counterintuitive to think that it is a good idea to share your customers with competitors. But decades of experience in direct marketing prove that it works big time.
I recently became aware of a next-generation model when our agency heard the pitch of Buysight, a start-up focused on leveraging co-op info plus other buy-point indicators. (Disclosure: Our agency went on to do a small project for the company, but it is no longer a client.)
Buysight identifies consumers who demonstrate many overt buying behaviors (e.g., visiting online stores, shopping comparison sites) and then makes these prospects available with a marketplace model. It uses what it calls buyer intent mapping to demonstrate bona fide in-market status. This targeting approach, coupled with cost-per-click buying and item-level dynamic creative, round out the company's offering.
Buysight's co-op approach works such that an advertiser gets first crack at its site visitors. But if the site presents them with a number of retargeting messages and they don't respond, then other brands are given a crack at them. Thus, the retailer doesn't suffer by sharing users, but can benefit from being able to target other retailers' non-responders.
This year, search retargeting appears to have "gone mainstream." Search retargeters -- companies like Magnetic and Chango -- collect data related to what individuals search for. By targeting banners with this data, these providers extend near-SEM-level results into display.
Why would you buy banners tied to searches instead of just bidding on Google keywords? A few reasons:
- First, there is only so much search result inventory out there. Its incidence is controlled by consumer behavior. I cannot will consumers to search for "refi info." I can only participate in the keyword auction for when they do.
- Second, only one brand gets to be first in SEM. And only a few brands get to appear in results. I just searched for "mortgage refinance" on Google. The only bona fide bank that bid enough for my eyeballs was Bank of America. The other seven results were aggregators and loan companies. Assuming that Wells Fargo, Chase, Citi, and other banks want my business, they could execute search retargeting and show me banners.
- Third, the cost of search words is climbing rapidly. Costs in many categories have reached beyond a level at which it is economic for them to purchase the keywords.
- Fourth, just because I searched for "mortgage refinance" doesn't mean I clicked on a result and applied. Search retargeting enables brands to connect with me as I consider my options.
That's why you would tie your banners to searches. The fact that search retargeting focuses on a set of highly predictive bottom-of-the-funnel touch points means that certain brands are already dialing their phones.
Another interesting approach comes from Owner IQ, which calls its model "owner targeting." The company collects data on user actions that demonstrate product ownership to deliver prospects with highly probable interest in categories, brands, and interests. Then it delivers media as an ad network. Like regular networks, it measures interest via site visit behavioral data, but adds ownership data points to sharpen its perspective.
Online actions that demonstrate ownership might include:
- Looking at the online owner's manual for a TV set
- Analysis of searches and page visits for specific items (e.g., a search for replacement iPod accessories would confirm ownership of an iPod)
- Product registrations online
- Actual purchase data from online retailers
You might think that knowing what people have already purchased defeats the purpose of targeting. After all, we want to sell stuff. But by understanding someone's ownerships, and adding that to other behaviors, Owner IQ says its model is highly predictive of future purchases. It also tracks "duress behaviors" that are predictive of the desire to trade up from existing items.
Other brands are getting results from these tactics, in part because each represents a "green field" of potential insight. That's not to diminish the relevance of regular BT data, but rather to suggest that lots of brands can benefit from more approaches and data sets. Marketers often despair that they cannot identify "on buttons" for their businesses -- paths to better results. Fortunately for those of us in digital, these five approaches represent insights and information that might well take us to Spinal Tap's Holy Grail of "11."
You are a heck of a lot more likely to get performance improvements from new data.
On Twitter? Follow iMedia Connection at @iMediaTweet.