ellipsis flag icon-blogicon-check icon-comments icon-email icon-error icon-facebook icon-follow-comment icon-googleicon-hamburger icon-imedia-blog icon-imediaicon-instagramicon-left-arrow icon-linked-in icon-linked icon-linkedin icon-multi-page-view icon-person icon-print icon-right-arrow icon-save icon-searchicon-share-arrow icon-single-page-view icon-tag icon-twitter icon-unfollow icon-upload icon-valid icon-video-play icon-views icon-website icon-youtubelogo-imedia-white logo-imedia logo-mediaWhite review-star thumbs_down thumbs_up

3 offline tools that can teach us about online data

Craig A.D. Dillon
3 offline tools that can teach us about online data Craig A.D. Dillon
VIEW SINGLE PAGE


Online marketing companies are utilizing the latest technology to attract, keep, identify, and select the best customers, right? Well, not exactly. While most online marketing companies utilize modern delivery technology, most are using little or no analytics technology. In the rare cases when they do, it is typically technology developed in the offline world in the 1970s and 1980s. Online marketing companies have 40 years of analytics technology to catch up on and, just as what happened in the offline world, those online companies that focus on analytics will dominate their competition.


There are three notable companies innovating data and predictive analytics that can serve as examples of how this technology can be applied to the online world. These innovations are being used to provide fraud detection strategies, predict consumer behaviors, and identify and target customers, resulting in more highly attuned customer interactions. In this article we look at what they are doing, how it works and what the business benefits can be for online marketing.


Fraud detection strategies
In any pay-per-click scenario, knowing how to detect fraud is crucial since it is so easy to manually or technologically click on links without intent to buy. Because advertisers pay for clicks, whether honestly clicked or not, service providers are obligated to protect advertisers from this kind of threat.


This online problem has been addressed through the innovations of HNC Software (now part of FICO), a little-known company that figured out how to predict possible credit card fraud. HNC's statisticians developed mathematical profiles of each credit card customer's behavior. They then built predictive models utilizing neural network technology (mathematically mimics biological neurons) to predict when a given card's behavior is abnormal or when it starts to look like known fraudulent behavior. These models have been under constant development for 20 years and are so accurate that they can often flag a card as stolen within the first or second transaction. HNC's solution, called Falcon, is now universally used around the world.


By using analytics in a manner similar to the way HNC Software predicts fraud on credit card transactions, online companies can predict and detect fraud based on historical and current click patterns, helping to ensure that the clicks advertisers pay for are the clicks that come from honest prospects.


Fraud is also a potential problem for cost-per-action (CPA) businesses, where advertisers pay based on the creation of a lead. In this business model, lower quality leads become trouble for advertisers when consumers charge back purchases. Here again, smart online companies can apply advanced analytical technologies utilizing known fraud patterns from the past to predict fraud today.


Behavior prediction
The power of analytics goes beyond spotlighting fraudulent activity to actually predicting behaviors, and as a result, helping to improve online response rates. Bill Fair and Earl Isaac revolutionized the credit systems by creating an algorithm that could, with great accuracy, predict the risk of lending to a consumer based on their prior credit history. This algorithm became the basis of the FICO score, and today it's the standard mechanism for assessing credit risk in the U.S., essentially predicting a consumer's behavior when it comes to paying back a debt. 


Using predictive analytic technology similar to that developed by Fair and Isaac, online businesses can predict what their customers may respond to or what advertisements may be most effective. In the FICO score case, consumers' past loan payment behavior is a powerful predictor of their ability to service current loans. Similarly, in the online marketing environment, how an individual responds to online ads, supplemented by other information about current interests, life stage, or geographic location, can effectively predict what advertisement to present today.


Next page >>

If the outcome decision in the FICO example is changed from "is this person a good credit risk?" to "is this person interested in my product?" then you have an analytical match. Recently developed predictive models can identify the best product to offer each consumer online. Results indicate that it is possible to improve click-through rates by 75 percent.


Online companies can implement numerous data-driven analytical techniques to help advertisers and publishers improve advertising response rates. This type of analysis creates a highly targeted offer that meets a current consumer need and is likely to motivate a response. Predictive models are just the beginning of the analytics technology roadmap that can be applied to this problem.


Customer identification and targeting
The online advertising industry has many of the same challenges addressed in the offline advertising industry three decades ago. Knowing where your customer is doing business is crucial. For example, an individual visits your website but never identifies himself/herself. How do you know whether or not they are someone your company has interacted with in the past? Further, suppose your online company could, at a minimum, identify that it had a relationship with this consumer; how does your company know what that consumer might be interested in?


Determining these relationships is essential for effective marketing and ensuring the consumer gets relevant offers. That's the kind of analysis that comes from the data and analytics innovator Acxiom.


Acxiom uses applied analytics, the science of codifying the meanings hidden within vast quantities of data, to recognize consumers. Specifically, organizations today may have consumer relationships that span many years and many departments, such as a car loan, a credit card, and perhaps a mortgage. The history of those relationships probably resides in multiple databases across the enterprise.


So in a typical organization, there is a high probability the company has a record of many individual relationships with, say, John Smith. It is often difficult to discern one John Smith from another John Smith and perhaps more importantly, to group the same John Smiths together. Using historical data and sophisticated analytics, companies now can make these determinations.


Acxiom's technology also creates a consumer segmentation system that works for the online world. This segmentation system groups various demographic characteristics at the household level so products and services could be better targeted based on lifestyle and life stage.
 
For example, on any street there may be individuals who recently had children, a couple who are grandparents, others who recently retired, and others with kids in college. Clearly, each household on this street is at a different life stage and will likely respond very differently to various marketing offers. Acxiom's solutions allowed traditional advertisers to ensure they group all the relationships they have with a consumer together and ensure they properly understand that consumer so advertisements are targeted appropriately.


Today, these kinds of innovations are helping online companies bring well-targeted offers directly to the right consumers, analytically determining that every consumer interaction aligns the offer with the highest probability of response.


Conclusion
Over the past 30 years, Acxiom, HNC, and FICO innovated using data and analytics and then applied this intelligence in a manner that addressed fraud, risk, and marketing challenges. It is just as important to build this type of intelligence into the world of online advertising so that marketers can identify responsive customers, and so that consumers see offers they are interested in. Adoption of analytics technology in the offline world often propelled companies to dominate their markets, and the same will happen in the online world.


Craig A.D. Dillon is senior vice president of analytics and media buying at Inuvo


On Twitter? Follow iMedia at @iMediaTweet.

Comments

to leave comments.

Commenter: Rajesh Vinaykyaa

2010, October 29

I agree Craig, except PPC which i have discussed with many of my clients. They spent $$$$$$'000 all these years but had little success. You right that fraud detection is important but how many do this or have those experts to handle this.

Visha Consultants