Predict Your Audience’s Preferences, Digging into [x+1]

[x+1]‘s VP of product development, Howard Fiderer explains in iMediaConnection how to make consumer data actionable so that you can tailor users’ experiences to their tastes on their very first visit in Predict Your Audience’s Preferences.


For those of you who are unfamiliar with the company, [x+1] is the former Poindexter.  Originally a primary ad serving company, Poindexter raised an eight-figure round of private financing in March of 2005 and used the funds to re-define the company from product, to target market, to brand.  Starting with adapting the mathematical formula x+1, Poindexter reset itself in the marketplace.  “We’re using this as an opportunity to mark a stake in the ground for a category we’re trying to define, marketing optimization,” the company’s CEO Toby Gabriner told ClickZ News in April 2005.  [x+1] proposed to focus on advertiser-marketing within the advertiser’s site while maintaining their advertising business, or ad serving business.  Ideally, they saw the opportunity to integrate the two products into a prospect-drawing and customer-targeting model as the ideal go-to-market.


First problem to overcome, Poindexter’s ad server was not viewed as a tier-one ad server in the market.  Although it had respectable market share, their reporting capabilities were commonly considered to be sub-par.  Major advertisers like American Express, who had termed contracts with Poindexter, complained about the lack of report diversity and the limitations of data availability.  Other agency users also had issues with their reporting, when their advertiser clients forced Poindexter on them.  That was the model for Poindexter, they sold to the advertiser so the agency would have to use them. 


Actually, it was more of a resultant model.  Poindexter sold to advertisers because they were looking to sell their ad serving and their developing predictive customer targeting back-end solution.  They promised improved reporting but it was not coming fruition on the ad serving side for customers. 


A Perfect example with regard to reporting shortfalls is the concept of ‘view-through.’  A view-through is when a user sees an advertisement served by an ad server but does not click on it.  Later that user visits the desired landing page which is tagged by the ad server and can be measured back by the ad server as having seen that particular ad on the associated site (placement) where it has been displayed.  This is known throughout the industry as a ‘view-through’ and [x+1] can not measure it, or at least does not report on it to its advertising customers.


But what [x+1] was particularly good at was/is the site-side customer analytics and applying those anaytics for targeting.  Like Howard’s article describes.  This is what secured their relationships with large advertisers who were using their ad serving as well.  The conversion from Poindexter to [x+1] was incredibly intelligent because it was a migration toward their core competency.  Following the April announcement, rumors spread that [x+1] would be abandoning their ad serving business altogether.  However they have maintained their media+1 product line, which is their ad server.  With that, however, they are hardly ever encountered as a competitive bidder in the ad serving sales arena.


One likeable aspect of Howard’s article was that it was not self-promoting.  Howard opens the door to site-side predictive modeling and website customer conversion and retention but he kept it very high level.  In fact, it would have been nice if he had gotten more granular for us so that we could have a better understanding as to how to apply his concepts.  I never have a problem when people mention providers – even their own companies – so long as they mention competitors and highlight the best solutions without bias. 


[x+1] offers up two primary products: media+1 and site+1.  Howard’s article is focused on a capability delivered by the latter, a tool that matches messages and offers with audience segments to simplify and optimize visitor acquisition, enable a site to up-sell conversions and promote customer retention.  This of course is according to the [x+1] web site.


My experience, and the feedback that I have received from clients is that site+1 is [x+1] true wheelhouse offering.  As I have described, this is where their ability to enable an advertiser to confidently target excels.  Ad serving is a secondary competency.  In fact, [x+1] has partnered with ad servers like DoubleClick to allow an advertiser to take advantage of site+1 while working with another ad server.  If not already, I would expect integration with more ad servers to come.  Clearly the company respects to obvious stats.  The first is that people are not going to change ad servers to utilize site+1 – they are not going to adapt media+1, known to be inferior, just to have the ability to utilize site+1.  Secondly, if someone is already using an ad server – and DoubleClick represents like 50% of the market (good first partner to choose), then better to enable integration to open up a new customer base than to compete.


Reactionary with Insight

_uacct = “UA-980395-1″;

Gaining Ground with Behavioral Targeting: Mediaplex tells (Almost) All

So I have received a lot of email and even a couple of phone calls regarding yesterday’s article by Mediaplex’s Sean Quick in iMediaConnection.  How could I let an article on Gaining Ground with Behavioral Targeting go by without so much as a comment, when so much of my blog addresses this topic with such conviction?  Well, yesterday was my birthday so I was out having a much needed break!  But I’ll have at it now!


Okay, first of all I have to start off by saying that Sean did a good job of breaking up BT into two distinct categories – passive and active.  He used these descriptions to help us understand the difference between the classic event-based targeting capabilities of a TACODA and the re-targeting capabilities that I have been evangelizing on this blog.  I was so pleased to see Sean use the term customer re-targeting as well, since it was not an industry term when I first started this blog and my Urchin reports show Mediaplex on my blog everyday.  I take Sean’s usage of the term as a compliment J.


Recently I posted When to Blog and When to Publish, commenting on what I believe to be the hairline-thick line between what is informational and what is self-promotional publishing online.  I issued a request for feedback as to whether I should be blogging or publishing and if there are topics on my blog suitable for publishing, should you believe I should be publishing.  I was contacted by Brad Berens at iMediaConnection who suggested that I could follow the thought leadership track when writing for iMediaConnaction (with their help) and that my self-promotion and subjective style could be reserved for the blog.  Works for me!  Anyway, stay tuned as you may see some writing show up out there as well…


Anyway, back to Sean’s article.  While Sean does specifically highlight ValueClick’s Mediaplex as the provider of re-targeting, he also mentioned DoubleClick and  So I don’t fault him for direct self-promotion.  But what I do believe Sean fails to do is accurately depicting customer re-targeting as it has come to be defined. 


If anything I believe that what Sean has done is introduced a third form of behavioral targeting: (1) passive, (2) active event-based and (3) active segment-based.


Passive BT goes back to the event-based targeting that I have already discussed in many postings.  Sean describes it as follows:


Passive BT — also sometimes called Targeted Segments and other names — is generally done either through applications that reside on a user’s computer, such as downloaded software, or through tracking tags that reside on publishers’ websites.  In either case, these technologies anonymously record consumer web browsing activity.  The consumer is unaware that such tracking is occurring, as it doesn’t affect their surfing activity in any noticeable way.


The tracking information is collected and analyzed, and the cornerstone of this approach is the subsequent attempt to make increasingly educated guesses about a consumer’s interests based on the data in order to deliver timely and relevant marketing communications.


Examples of providers of this approach include Tacoda, Revenue Science and Claria.


The Active BT that Sean describes is “also called User Retargeting [and] consists of anonymously registering consumers’ proactive, direct interaction with a company’s marketing efforts. For example, display or email advertisements or material on the client’s website and then implementing follow-up marketing programs that address that explicitly expressed interest in an attempt to deepen the relationship and lead to conversion.”


So this is still event-based.  A better description is Event-Based Active BT.  The information leveraged for re-targeting is based on other marketing experiences.  Sean does not describe the application of this BT specifically in terms of online advertising, nor does he describe it in terms of preventing the re-prospecting of customers through online advertising.  In fact, what he is doing is explaining that Event-Based Active BT can coordinate disparate forms of online marketing to collectively create a concrete BT model.  Moreover, and what is entirely left out of this article is the PROCESS.


One thing that I am so careful to do in all of my postings – and what I believe will be a focus of articles that I would write for IMC – is to illuminate the procedural differences between various kinds of technologies so that people can come to distinguish them.  It is important to understand the impact of a first party design and a third party design for example.  The benefits and differences of these kinds of technologies transcend the advertising experience for both the advertiser and user – in fact for the publisher too. 


TruEffect holds the patent-pending rights to first party ad serving.  So if another ad server were to implement a first party design, they would violate that patent and would be putting their clients in a nefarious situation that would result in problems down the road.  Aside from that, other ad servers are not doing it that way anyway right now.  They have their own design using a third party cookie and a synchronization process.  They still have their ‘control the data and you own the client’ model.


Listen, Sean is not going to talk to me about it, obviously.  But maybe he will talk to you and then you can come back and talk to all of us on the blog.  OR, maybe Sean would like to come on here and have a discussion with all of us.  It would be great to bring the truth out.


If Mediaplex is using a third party cookie to conduct user re-targeting, it is historically synchronizing.  Latency comes into play and there are limitations to the benefits that only a real-time capability can bring to the table.  Only a first party cookie foundation is capable of doing it in real-time. 


If Mediaplex is having clients share record information so that Mediaplex can assign cookie values to people when they transact or otherwise experience a marketing event, it is still a third party cookie, foreign to the advertiser and so the limitations include:


  1. The ad serving data is mediaplex’s data, accessible only by mediaplex;
  2. The cookie information is not accessible by the advertiser;
  3. The ad serving information is not readily integratable with other technologies such as site-side analytics which may be another third party cookie (omniture) or could be a first party cookie (WebSideStory or Webtrends); and
  4. Mediaplex can not adjust targeting strategies in real-time, targets must be determined in advanced.

As we have covered in many entries on this blog.  First party ad serving, using a first party cookie, allows all of the four aforementioned limitations to be mitigated.  Most of all, targeting can happen in real-time.  An advertiser can change a cookie value, login to the ad server and change the targeting reaction to the cookie; and the change is instantaneous. 


With a third-party implementation, the advertiser has have to share the altered customer information with the ad server (Mediaplex, DoubleClick) and then the ad server has to start writing new cookies, which have to propagate, and then the ad targeting can begin.  Can you say latency?


Two other limitation issues: (1) SOX and (2) third party cookie deletion.


When a third party is handling your data, and you have SOX compliance issues, you have a potential problem.  Using a third-party cookie and a third party ad server, deploying event-based Active BT, means your data about your customers is being shared with a third party who is subsequently developing additional information about your customers and gate-keeping your access to that data.  You need to make sure that you have controls written about the handling of that data because it is out of your control.  This is not an issue with first party ad serving because all of your ad serving data flows directly through to the advertiser and is not withheld by the ad server.


Secondly, third party cookies get deleted over 40% of the time – Jupiter Research.  So only 60% of the Event-Based Active BT will be effective whereas over 90% of first party cookies are persistent.  Do the math and you will realize that leveraging a first party cookie will bring a much higher yield in re-targeting activities.


So Sean’s article stimulates the interest and probably results in some genuine leads to Mediaplex, but I wonder how far down the path you will get with Mediaplex before you come to realize that you have not come to engage with customer re-targeting but more event-based BT?


The third form of BT that I characterized earlier is Active Customer Re-targeting.  The fundamental distinguishing difference is that the advertiser is creating customer segments and is cookieing their customers directly, as opposed to the ad server cooking the customer.  Advertisers may cookie their customers through eCRM, eCommerce cycles, email processes, site-side analytic platforms (i.e., WebSideStory), landing pages (i.e., CoreMetrics) all using their own first party cookie.  They can create customer segment profiles that associate a user with a customer type, just like they do offline with direct mail, cataloging and telemarketing and then deploy customer re-targeting with their online advertising.


So Sean, it would be very interesting to have a follow-up article with which we hear how Mediaplex, and if you’re so inclined to research your other examples – and DoubleClick – conduct the Event-Based Active User Re-targeting.  But that might not be on your agenda.  Hey Brad, maybe I’ll write that article for you!


Reactionary with Insight.

_uacct = “UA-980395-1″;

When Will We Deliver On The Promise, Deliverying the Right Ad At the Right Time

In MediaPost’s Online Spin, Dave Morgan wrote When Will We Deliver On The Promise?.  He discussed a question that he received while sitting on a panel at the  DLD (Digital, Life, Design) conference in Munich. 

Walt Mossberg, personal technology columnist for The Wall Street Journal wanted to know: “When will the online advertising industry actually deliver on its outstanding promise of the last decade to present the right ad to the right person at the right time? When will we finally deliver a consumer experience where the vast majority of ads that we deliver are meaningful, and not junk and clutter on Web pages?”

Dave’s observation was that “the online ad industry is always touting its extraordinary technical capabilities to target relevant ads … [but] we are still giving consumers a terrible experience when it comes to the vast majority of ads that we place on their Web pages.”

Dave’s article argued the evolution of the industry and compared it to the evolution of print and broadcast.  But what he failed to do was point out the capabilities in the industry that do deliver on the promise of delivering the right ad to the right person at the right time.  As the CEO of TACODA, I was surprised not to hear him plug it, but then again his insight expands beyond self-promotion and so it was healthy to see him stay on topic.  Regardless, here is the response that I posted on his blog:

Solid arguments Dave, and a very positive outlook. I think to see the conformity trend makes a lot of sense. But I also think that repurposing existing technologies presents the opportunity to put the right ad in front of the right person now. I don’t know how you answered Walk Mossberg’s question, but he is not alone in that inquiry. It is a popular question and one that I know TACODA aims to answer with BT. Putting the right ad in front of the right person at the right time is the golden nugget. Knowing how to recognize the right person is the major task. BT does it with the prospecting of anonymous individuals as we know. Customer re-targeting can establish the same capability through first party ad serving as I have argued in the past. Together these two technologies can enable an advertiser to put the right message in front of prospects and customers while advertising on the web in real-time. The composition of the advertising audience can be measured so that customers no longer get re-prospected with promotional messages and customers get driven towward recurring revenue-minded actions. Putting the right message in front of the right person at the right time makes the advertising relevant.

Reactionary with Insight.

_uacct = “UA-980395-1″;

The First Party Ad Serving Rabbit Hole

Many customers who first begin to investigate DirectServe technology and the value proposition of customer re-targeting get that “wow” moment in their eyes when I speak with them.  People who read this blog contact me to learn more about the fundamental differences between traditional event-based behavioral targeting and customer-centric re-marketing.  When they get the premise of recognizing someone you know online through an ad campaign, gaining the capability to distinguish them based on a set of parameters that you set so that you can message to them uniquely, the “wow” expression rears its pretty face.


The value proposition conversation is what follows.


  1. No longer prospecting existing customers.  With first party ad serving, customers and prospects are distinguished so that you don’t prospect your existing customers but rather cross-promote products and up-sell to them, increasing recurring revenue opportunities


  1. Insight into Advertising Audience composition. For the first time, DirectServe enables an advertiser to measure the composition of their advertising audience – what % of the audience reach is comprised of existing customers and what % is comprised of prospects.  Being able to identify when it is time to move on from a site if prospecting is the primary objective becomes easy to quantify with first party ad serving.  Determining that a specific site is ripe for re-targeting is easy to determine as well if you find that the audience is largely comprised of your repeat-visiting customers.


  1. Target customer segment profiles that you set.  With first party ad serving, the cookie’s targeted by the ad server are set by the advertiser and are correlated to the advertiser’s customer segmentation model.  As that model grows, expands or evolves so too does the targeting reach of the ad server.  The ad serve does not need to even know the meaning of the cookie value, only what creative group to associate with a particular cookie value.  The advertiser maintains complete control over their messaging strategy.  With third party implementations of behavioral targeting – like the one referenced in Tom Hepos’ recent iMediaConnection article – the third party ad server would be keenly aware of the value and meaning of the cookie targets which exposes customer knowledge to a third party.  This can open Sarbanes Oxley doors or other privacy issues.


  1. Write ad serving data to your cookie that is relevant to your site.  With first party ad serving the ad server can write to your first party cookie values that reflect banners and site sequences that reflect that acquisition marketing campaign that ultimately produced the visit.  This data can be integrated with site-side analytics software or CRM solutions to create an entirely new class of customer profile data.


Now back to the rabbit hole.  I’ve talked about the 80/20 rule.  With DirectServe its more like 90/10.  Most of our clients engage DirectServe for re-targeting.  They write cookies reflecting customer segment values and we target those values with unique banners intended to reach those customer segment groups.  These clients are most frequently represented by ad agencies who do not look to expand the DirectServe capability into the data realm, and who don’t need to.  The benefits of DirectServe re-targeting are huge and so that is why we focus so heavily on it with out GTM.


The rabbit hole for DirectServe has to do with both Re-Targeting and the rest of DirectServe, what we can Analytcs.  From here on out I will just talk about the Re-targeting.


When you look at a traditional online ad campaign, you buy media across X number of sites with Y number of banner sizes to create Z number of placements represented by banner size and site-sitesection combinations.  With me?  10 sites with 2 different sizes demands a minimum of 2 creative.  But to optimize a campaign you will likely have three creative in rotation, so that’s 6 creative total to start. 


With DirectServe you will be targeting customer segments in addition to prospecting.  So the campaign plan described above is appended by the following:


If there are 3 customer segments (High Value Shopper, Low Value Shopper, Has not shopped in 6 mos), than there is the need to create 3 messages to each of these groups as well just like with the prospecting campaign so that you can optimize the campaign.  So that is 3 creative x 3 segments = 9 creative.  Add that to the prospecting campaign and you are up to 15 creative for just the two sizes that you will be running on the 10 sites that you have bought advertising on.


In summary:


Prospect Campaign:

·          Messaging – 3 creative

·          2 creative sizes

·          6 creative total

Re-Targeting Campaign:

·          Messaging – 3 messages x 3 groups = 9

·          2 creative sizes

·          18 creative total

Campaign Total:

·          12 unique messages

·          24 unique creative

·          (10 sites, 2 sizes=20 spots) = 240 placements


In the end, DirectServe introduces a great opportunity to dissect your audience up so that you can message exclusively to different customer segments and drive home specific results.  But the planning and creative requirements need to be kept in check too.  Out clients start off small on the level of something like the example described about just to begin to experience impact – results that are impressive and significant.  Then they begin to determine and test whether multiple customer segments is where they want to diversify or creative messaging or a combination of both.  The budget drain on message planning and creative is the variable that requires management with DirectServe.  But we have clients that are seeing 15-20% improvement in recurring revenue opportunities and that more than justifies the cost of additional creative!

_uacct = “UA-980395-1″;