Commentary

Delivering The Message

by Phil Leggiere , January 14, 2009

The more aggressively marketers "bid for relevancy" by leveraging ever more pinpoint modalities of micro-targeting, be it via behavioral and other flavors, the more imperative it becomes to actively deliver on the promise of relevancy, with truly personalized messaging. Yet, as Calvin Lui, President and CEO of Tumri, explains below, it's precisely in the serving of creative that advertising has fallen short, dashing much of the potential of enhanced targeting.

Behavioral Insider: What is the relationship between dynamic ad serving, as Tumri is proposing it, and enhanced approaches to display ad targeting?

Calvin Lui: We're moving from a world where advertisers target by programming content (which is just an extension of traditional print and TV) to one where we can dig deeper both into people using the media, and how they're using it. Over the years we've gained a much deeper knowledge into behavior, context and location. Now the question becomes 'How do you adjust and adapt your message to capitalize on that enhanced knowledge?'

Most marketers have crossed that threshold into micro-targeting. The challenge is that for all the insight they cull from their targeting platforms, they are hamstrung when it comes to effectively adjusting their message, in all dimensions, to really take advantage of what they know about their audiences. Advertisers need to adjust the message to the contact; otherwise you haven't unleashed the power of that targeting.

BI: Could you describe how the modular approach to ad creation and delivery Tumri has developed works?

Lui: Traditionally, display ads were constructed by building self-contained, individual ad units that were loaded separately onto a server. If you were Coca-Cola, for example, you may create an individual ad geared to males, another to females. You'd also have one developed for the Chicago market and another for New York, each one a self-contained file and then the ad server would select which ad to run.

So the premise of Tumri's AdPod has been to enable marketers to more effectively deploy ad content by taking a template and breaking down their ads into different subcomponents, be it the headline, the product, call to action, pricing, background, and/or particular images. Each subcomponent of that template can be changed on-the-fly depending on who is viewing the ad, and what the context is, so that in addition to the ad placement the messaging and creative become adaptable as well. If Ford shows an ad to Sally the Soccer Mom in Indiana the ad may feature an SUV and the messaging can stress roominess and safety. Eddie the surfer would get an ad for a convertible and the ad would have a beach scene in the background and would stress a cash-back discount.

BI: I know one key partnership you've announced recently has been with Google's contextual network. What is the nature of the distinction you make between contextual targeting as traditionally understood and context-aware targeting -- and how does that play out in actual practice?

Lui: What we're doing with Google is that for the first time they're opening up the interface on their contextual network. So as an ad is being served they pass us keyword information and we adjust the ad subcomponents in real time, based on the context of the page the reader is looking at on a keyword level. Examining the contextual information and marrying that with past search and behavioral patterns, elevates the level of targeting.

So how does this relate to the Google content network? If you use an ad exchange like ContextWeb or folks like Yahoo for your ad serving, the algorithm weighs the content on the page (headline, keywords, phrases, etc.) to deliver a relevant ad. With Tumri, Google is opening their interface and architecture to allow Tumri to access keywords for pages. When an ad is served, the Google content network will pass through Tumri with recommendation of content, and Tumri will refine that further.

BI: What are the benefits of this synergy to advertisers?

Lui: The obvious benefit is much greater relevancy in front of the consumer and a higher lift in response rates. Even more, there's lower cost of production. In the traditional paradigm, Coca-Cola or Ford would need dozens of versions of hard assets. For example, if there's an attempt to geo-target 30 DMAs, a male and female demographic, and numerous sub-demographics within each category. Then you have three different messages and five different format sizes. You put that all together and do the math, [and] you have hundreds, if not, thousands of possibilities. Way too many to leverage effectively, if you have to develop separate hard assets for each possible iteration.

Another added dimension of effectiveness is that with Tumri, reporting can be done based on subcomponent. You can test dozens of different elements very quickly and find out, for example, that Sally the Soccer Mom responds best to a unique mix of sub-components, say recipe number 15, promotion number 4 and background number 3, and then optimize your mix of subcomponents accordingly.

behavioral targeting

6 comments about "Delivering The Message".

Check to receive email when comments are posted.

John Berard from Credible Context, January 14, 2009 at 2:37 p.m.
Tumri is to be commended for its part in increasing advertising's "bid for relevancy." But deploying increasingly complex hypotheses -- combining context, behavior and location then sprinkling a bit of optimization -- is still a guess.
I have come to think that the best way to deliver relevant advertising is to ask people their interests and let them manage that portfolio over time. Publishers and advertisers can rely on such pre-qualification to deliver relevance.
Radical, but rooted in the best traditions of successful retailing.
Reply

Joe Fredericks, January 14, 2009 at 3:58 p.m.

You guys make it sound like rocket science. Is that your modality or are you just happy to see me?

Multi-variate ads make sense. Another way to solve the ROI puzzle for the marketer.

The End

Monica Bower from TERiX Computer Service, January 15, 2009 at 10:33 a.m.

The unfortunate truth is that many advertisers are much better at metrics than they are at advertising. To be more specific, the numbers are the easy part - the hard part is creating a relevant message, especially when your actual product - shoes, soda, lawncare - is generally irrelevant to everyone outside of the very small subset of people who are actively engaged in looking for something like that already (in which case the ad is irrelevant, and all that matters is your SEO, right?) Increasingly accurate usage and redemption data points a finger back at the retailers - not the agency message-crafters but their clients - as being wholly unprepared to do anything other than their normal daily routine, meaning they have no effective way to leverage any of the usage and redemption data. There are some exceptions, but the rule is that a campaign can be no more effective than the client is capable of managing, and what they think they can do is usually orders of magnitude more than what their feet on the street can actually do in terms of properly servicing a branded campaign. The guy at the counter or in the cubicle is usually totally unaware of what marketing or the agency is up to.

Scott Milener from AdRocket, Inc., January 15, 2009 at 7:42 p.m.

I think Tumri is on the right path, whether the surfboard ad vs. the snow ad makes a difference, maybe. It certainly would if they actually knew the girl is a surfer, hard to get that data.

Mainly the benefit here is in the rapid a/b testing and accurate reporting of what's working. Its hard to move quickly without those tools.

AdRocket gets lift based on parameters similar to what Tumri is doing, but running text based ads or offers.

I agree that if you can get a consumer to tell you what they are interested in, great! But very hard to get that interaction.

Daniel Harrison, January 16, 2009 at 4:48 p.m.

Combining Behavioral Targeting with Dynamic Messaging enables the marketer to incorporate key info into the decision of which creative/message to dynamically build.

http://www.mediapost.com/publications/index.cfm?fa=Articles.showArticleHomePage&art_aid=97346

Pascal Bensoussan from Aggregate Knowledge, January 21, 2009 at 1:48 a.m.

John makes a good point.

A great example of very effective behavioral targeting is retargeting. A consumer visits a site, states his or her interests more or less precisely and leaves without converting. This scenario presents a great opportunity to re-engage with the consumer based on their past behavior. Some ad serving companies simply present a simple generic creative that might hit or miss its target. Other ad serving companies might display the product that was left in the cart or a selection of popular products. Those techniques have proven that retargeting works.

At Aggregate Knowledge, we help advertisers re-engage with past visitors, from home page and landing page visitors to cart abandoners, with relevant and personalized product recommendations. We can generate millions of creative variations with a single tag and a single creative template.

We have experimented with different scenarios and successfully brought back lost landing page visitors with hot and popular products and offers, re-engaged category browsers with personalized recommendations, and converted cart abandoners with alternative products or up-sells and promotional messages.