Commentary

Does Online Privacy Legislation Mean a Substantial Shift in the Interactive Media Landscape?

With the early May introduction of the Boucher-Stearns Online Privacy Act, a bipartisan draft legislation poised to regulate personally identifiable information transmitted online, marketers focused in interactive media may be facing an evolving informational landscape with regard to their ability to target users on the Web. While privacy advocacy groups have unanimously panned the draft version of the bill as toothless, online media trade groups like the IAB, favoring self-regulation, still have questions and concerns.

For example, current language in the bill leaves room for interpretation regarding what types of information are considered 'Personally Identifiable.' While things like medical information and Social Security numbers clearly fall into the categories of both 'Personal' and 'Identifiable', warranting close regulation, others -- for example, cookies on a user's browser, or a user's IP address -- are less obvious. For example, what if a user reads an article about Type 2 Diabetes on a popular online health site, and is then cookied as eligible for targeting for products like blood sugar monitors? Is that medical information? Should the user have to opt-in to be targeted for ads in that manner?

As somewhat of a positive sign for the industry, the current draft does include some special treatment for user information passed between individual Web sites and Ad Networks. This process, more or less the exchange of cookies for the purpose of behavioral and demographic targeting, is a fundamental piece of audience segmentation on the Web as it stands today. While the bill states that most personally identifiable information can only be passed when a user 'opts in', an exception is made for advertising companies exchanging cookies, as long as there is a "clear, easy to find link... that allows users to ... opt out," and users are notified that they are being targeted.

However, if pressure from privacy groups increases, the possibility exists that this fundamental element of online targeting -- the transaction of cookies from one company to another -- will be concretely defined as pertaining to personally identifiable information, thus requiring 'opt-in' consent from users and nullifying the exemption clause currently in the legislation. In other words, people would have to volunteer information on their online behavior, demographic information, and geographic location to advertisers trying to market to them.

If this legislation were to be enacted in this stringent a fashion, what are some possible outcomes for the Online Media landscape as it stands?

Vendors who rely on Third Party Data to define their product will be challenged.

The groups facing the biggest challenge would be those that have the most invested in buying and selling third-party data. Sellers of the data, like Acxiom, BlueKai, and eXelate, have built whole portions of their businesses creating and aggregating cookies based upon user behavior. In the future, they may have to seriously redefine their services and identity. As the sheer volume of user data would likely diminish, the market will likely see some consolidation and partnership among remaining players.

The other group rethinking their business plan would be providers of DSPs (Demand Side Platforms), second-tier ad networks, and essentially anyone else relying on exchanges to build their inventory pool. Without the ability to layer targeting data on top of network-level ad impressions, many would be left with groups of sites loosely bucketed by content similarity and of course, pure RON (Run of Network) offerings -- which for the record, tend to underperform. Ultimately, a 'thinning of the herd' would ensue, as it would be nearly impossible for these companies to differentiate themselves when pressed by media buyers.

Growth of Reciprocal Data Relationships
As of now, marketers have enjoyed the one-sided benefit of targeting users based upon a number of data points including age, gender, HHI, online behaviors, offline behaviors, and much more, compensating users solely with content. If the environment changes in such a way that users would have to be incentivized to give up information about themselves for the purposes of being advertised to, a new value exchange would need to be created.

One option that will likely appear early on is contesting and sweepstakes. Essentially, the pitch would be: "Fill out this form, we'll cookie you, and you'll be entered to win a trip to Hawaii." It's an old routine from the playbook. However, collecting data in this manner skews your targeting base significantly, because the demographic and psychographic characteristics of individuals who respond to those offers is largely different than the population as a whole. In other words, you'll be left with a list saturated by 'sweepstakes junkies.'

Another option, more fitting of the Web 2.0 environment, is the value exchange created by companies like Mint.com. The free online financial management tool (acquired by Intuit, the makers of Quicken, for $170 MM in 2009), collects sensitive financial data about its users in an opt-in environment. In return, the user gets free budgeting tools and money management advice. Where Mint makes its revenue, however, is in its recommendation of financial products like credit cards and insurance. To some, it might sound unsavory on paper, but there is something distinctly valuable in being pitched for a credit card that the software demonstrates will save you money over time, based upon what it knows about your spending habits and available interest rates. Same goes for checking accounts, savings accounts, auto insurance, IRAs, and more.

A third option, likely to take hold in sponsorships of longer-form content, is ad selection and user choice. Recently, a number of publishers and marketers have made a greater effort to provide more relevant advertising to their users. Hulu has integrated a 'is this ad relevant to you?' button that appears during commercial breaks, ideally to learn and later to increase relevancy. Also, Publicis agency Vivaki spearheaded Ad Selector, a more scalable platform that allows users to select product categories for advertising appearing in long-form content, in real-time.

Increased Focus on Contextual Targeting
There are a number of companies that already have a great head start on this methodology, and would benefit almost immediately from privacy legislation, as this type of targeting doesn't rely on information about the user. Rather, it uses semantic targeting technology, which can read the text on a page and make a guess as to what ads would be relevant to the content.

However, the space still has a great deal of room for innovation. An easy example that appeared a few years ago was 'Hilton, Paris.' The difference between hotels in France and a socialite with questionable values produces very different advertising opportunities, especially when evaluated by a computer, devoid of cultural context. While the methodologies have certainly become more sophisticated in recent years, computers still show signs of clumsiness on a regular basis. Two companies that have a considerable head start, however, are Google, with their Google Content Network, and ContextWeb. As more attention is shifted to this technology, more minds will be dedicated to refining the space, hopefully increasing its dependability to drive performance.

Growth of Vertical Network Segmentation and Emphasis on Content Adjacency
It is likely that advertisers would increasingly rely on traditional media segmentation models. This includes planning techniques used by most media professionals already, relying on ComScore and Nielsen @Plan, as well as site direct buys that are relevant to the product category.

In addition, the growth of Vertical Ad Networks, or collections of sites that are similar in content or (publisher perception of) audience, will continue an upward trend.

So as marketers, what do we do?

Like any good investor, you hedge your media budget. Do not rely solely on one technology to characterize your targeting. For example, the online team at Harmelin Media largely favors Channel Strategy, the marketing equivalent of putting your eggs in several baskets, and optimizing toward top performers. "Planning in this fashion allows us to not only maximize our chances for successful campaigns, but also increase the opportunity to learn on behalf of clients, and see what really works for a particular brand or message," says group VP Brad Bernard.

In addition, the industry as a whole may benefit by increasing emphasis on user education. Most people online are not aware of the ways they are currently being targeted, and furthermore, have not been exposed to the potential benefits. Many users may respond positively to the options to make their advertising more relevant, especially when faced with alternatives like pay walls.

7 comments about "Does Online Privacy Legislation Mean a Substantial Shift in the Interactive Media Landscape?".
Check to receive email when comments are posted.
  1. Donna Zelzer from Midwifery Today, June 18, 2010 at 10:31 a.m.

    ". For example, what if a user reads an article about Type 2 Diabetes on a popular online health site, and is then cookied as eligible for targeting for products like blood sugar monitors? Is that medical information? Should the user have to opt-in to be targeted for ads in that manner?"

    Yes, I think we should have to opt-in. For one thing, some one could be reading about Type 2 Diabetes without actually having Diabetes themselves or even know some one who has (research for an article, for example.) For another, why not just advertising the monitors on appropriate sites rather than following people around the internet trying to sell them things?

  2. Donna Zelzer from Midwifery Today, June 18, 2010 at 10:35 a.m.

    I meant to say "why not just advertise: - not fully awake yet!

  3. Aaron Reinitz from VivaKi, June 18, 2010 at 11:14 a.m.

    Good question Donna- the answer is scale. There are only so many content relevant sites out there that an advertiser can use in that manner. Further, the rates on those sites tend to be very high. A balance between relevance, impact, scale, and efficiency is key to a good media plan.

  4. David Kohlberg from Varick, June 18, 2010 at 11:22 a.m.

    Great article Aaron-

    Very well written and I thought it was very subjective. It will be a long time before this issue is resolved. In the meantime advertisers should take advantage best they can before anything changes.

  5. Jason Kaminsky, June 18, 2010 at 11:35 a.m.

    Really well written article Aaron. I think that we as consumers certainly make consessions for getting what we want when we want it. Part of tha bargain has become recieving targeted ads of course. On one end it feels more engaging to have relevant ads delivered to me, as opposed to those which have no relevance in my life. On the other end the question I always seem to ask myself is, will I be offended one day when a marketer sends me advertisements for incontinence pads or hair growth formula. That may be the line where the tracking and numbers game becomes too personal. It's a tricky line to ride.

  6. John Grono from GAP Research, June 18, 2010 at 10:14 p.m.

    Excellent article Aaron.

    I do have a question that relates to your reply to Donna's comment referring to scale. Given that there is a limited number of relevant sites (in the Type II diabetes example), isn't the rest just "spray and pray" wastage? Isn't this the self-same argument used against the traditional mass media such as television, press and radio? If I was a marketer I'd be taking the responsible and ethical high-ground of opt-in.

    Consideration in the privacy debate also needs to be given to the confluence of potentially personally identifiable information, rather than the individual bits of information. If a marketer knows that I am male, that is hardly targeted - I could be 1 in 3.3 billion people. If they know that I am 54 years old I could be 1 in around 100 million people. If they know I am in Australia I could be 1 in 22 million people. However, knowing all three pieces of information (all easily captured online) means I am 1 in 125,000 people on this planet. It's not too hard to 'scrape' my postcode/zipcode, and they'd have a fair idea of the industry I work in, and all of a sudden a pretty clear picture can be built.

  7. Aaron Reinitz from VivaKi, June 19, 2010 at 8:03 p.m.

    Hi John, thanks for commenting.

    To answer your first question, I think that most digital planners would agree with me that, while typically, endemically relevant placements do perform strongly, adding a layer of data-driven media (BT, Retargeting, etc) tends to drive efficiency and overall reach without entirely sacrificing performance. To put it another way, a popular health site may be very effective, but also 3-5 times more expensive than a Network or DSP using Behavioral Targeting to reach the very same user. The question then becomes, at what rate do the endemic site and network need to perform individually to justify their own cost? So in reality, it's not 'spray and pray' as you put it, more so 'test and allocate', utilizing sites and channels as role players to maximize investment for the client. That is the fundamental difference between planning online and planning in traditional media; the ability to scientifically evaluate media performance in real time.

    As far as your point of responsibility and ethics, yes, I do believe that the industry could do a better job of informing consumers just how their data is being used. However, ethically speaking, the purpose of this data usage is fundamentally valuable to the consumer: it provides them a more relevant ad experience. To put it in different terms, advertisers don't want to waste their media budgets, users don't want exposure to irrelevant advertising, and this data is a tool to satisfy both of those needs.

    Lastly, regarding your description of creating sophisticated targeting criteria, that's already done. Not sure if anyone from a Network or Data Provider would like to chime in here, but as a planner, I've been presented with several targeting products that can combine demographic, psychographic, and behavioral information in nearly any combination (within reason). Also, the data points you describe such as age, gender, and geographic location, by IAB standards, are not considered personally identifiable, as these datapoints are always defined within a range. For example, 'males, 35-54 in Australia' is targeting criteria, 'John Grono, 54 years old, 45 Walker street' is personally identifiable information.

    Aaron

Next story loading loading..