Commentary

Testing And Targeting: In Search of Synergy

Variety may be the spice of life, but in marketing -- as in the real world -- habit often limits options and possibilities far more than needed. To truly optimize the advantages of all the rich, diverse targeting data generated by customers online, marketers should expand their options by integrating targeted content with continuous real-time multi-variable testing, Mark Wachen, managing director of Optimost at Interwoven, explains below.

BI: Interwoven is pretty clear about not being a behavioral targeting firm per se, but rather an optimization company. Where does a behavioral component fit into the optimization mix in your new Adaptive Targeting platform?

 Wachen: The usually unconscious working assumption of targeting platforms has been that their particular 'special sauce' is best. The reality is that every piece of data, whether it's geographic or contextual or behavioral, is important -- and even more important when related to and integrated with other data. The particular levers may vary by advertiser or campaign, but all are potentially leveragable.

Targeting has remained a very siloed area. You have diverse options, geographic, day-parting, contextual, demographic, and behavioral, and subsections or varieties within each. But ultimately options have been limited in possible outputs.

Traditional Web site targeting uses the marketer's insight to match people with content using rules. Data mining allows offer optimization and behavioral targeting to cluster based on similar characteristics, but content is still limited to a set of fixed offers. Meaning that, if you're focusing on behavioral data, for instance, you have a very constrained set of options. You may be limited to only two or three offers to target. So, for any given customer prospect, you're forced to place a person in one of those three buckets whether they're really relevant or not.


BI: What's held back this integration?

 Wachen: The reason for this siloed approach, or at least a major reason, is that multivariable testing (MVT) and targeting have yet to be seamlessly integrated. We leverage multivariable testing and targeting with Interwoven Optimost Adaptive Targeting, so we can deal with hundreds and even thousands of possible permutations.

The most ideal situation for a marketer is to be able to continually test and tweak what and how you're targeting in real time. Testing without targeting is like peanut butter without jelly. So, what we've introduced is a system that evaluates every visitor and delivers targeted content that takes into account every data point we have about them, then conducting MVT for virtually unlimited content options.

BI: How will the better integration of testing and targeting change the way marketers view testing?

 Wachen: Most marketers continue to think of multivariable testing as optimizing offer A against offer B. That is certainly one important thing to be testing. But it's only one variable and there are many more. When somebody visits a page, they are engaged all the time in a series of small decisions. When they come to a form, they are judging whether it's too long to fill out or contains too many personal questions. They are making decisions as they read copy about whether what they have to read is too detailed or technical, or whether it's not detailed enough to suit their needs.

When visitors arrive, the system takes into account the context, the search keyword they used, the affiliate site they came from, their IP Address, cookies from past visits and the time of day, day of week, weekday vs. weekend, and other criteria. Beyond this, we can then overlay demographic data from third-party sources based on their zip code, or profile data from other customer record systems such as CRM and ERP.

From there we can learn from the browser about behavioral aspects such as where they go, what they click on, how long they stay on a particular page, what they download, and  visitor-captured data they enter to profiles,  trip planning details, or survey questions.


BI: How does this work in practice?

 Wachen: One Adaptive Targeting beta customer Interwoven worked with was America's Test Kitchen. They wanted to optimize their Web site to cross-sell. In their case, we started with multivariable testing and overlayed demographic data. We saw, for example, that audiences from areas that had larger concentrations of older adults responded significantly better to things like larger fonts and buttons, and shorter, simplified lead copy.

On the other hand, with other visitors those particular kinds of stylistic tweaks made no real difference either way in responsiveness.  This was an interesting revelation to the client. Once this new segment of 'retirees' was discovered to be using the Web site in critical numbers, whenever a visitor from the Geo IP addresses associated with this segment arrived at the site, the system could display the larger red buttons and larger font size for product pages of interest to this group. This could then be taken one step further, as targeted landing pages and even a microsite for this new segment of site visitors could be created.

Next story loading loading..