Laserlike

Today Yahoo! announced that it would join AOL in letting users opt-out of behaviorally targeted ads.  This is clearly a response to pressure from the United States Congress and the FTC.  As an article last year in The Washington Post discusses, this all started about one year ago when privacy advocates played on the success of the Do Not Call registry.  The European Union, not to be left out, has the Article 29 Working Party.

I was a big supporter of the Do Not Call registry.  Having advertisers interrupt a family dinner (or any other time, for that matter) is wrong.  There isn’t a benefit that covers the cost of the interruption, and the user did not have the option to avoid the cost.  In the online world, the parallel is email spam.  Advertising is not spam — consumers choose to pursue a benefit (content or services) and pay with their attention.  Consumers can choose not to visit a site.  If they do visit, they decide when to visit and can always choose not to visit again.  This is how most publishers pay their bills.  And greater targeting will allow for better and more services, which will ultimately benefit consumers.  

To say that you should let the consumer decide whether or not to receive behaviorally targeted ads is like saying that we should let all consumers decide whether or not to pay for products at a retail outlet.  

Advertising technology backgrounder.

Behavioral targeting (BT).

Behavioral targeting is technology which uses a user’s historic behavior in order to predict present interests for the purposes of advertisement targeting.  BT can increase the price that an advertiser will pay by an order of magnitude.

Content targeting.

Google prefers to use in-session behaviors to surface advertisements — for example, Gmail surfaces ads based on the content of the page you are viewing [you may recall the brouhaha over content indexing of Gmail several years ago].  

Search advertising.

There is no doubt that having a user tell you what they want right now and then serving up an ad alongside the desired content offers the best ad targeting in the world.  Unfortunately, for everyone other than Google, search is a natural monopoly.  

Why is behavioral targeting so important?

I would estimate that the revenue per thousand page views (RPM) of a search product is ~$50-$75, that the RPM of a highly targeted category with professional content (e.g., sports on ESPN or finance on CBS Marketwatch) is around $20-$30, and that the RPM for just about everything else (mail, social networks, user generated content sites) averages <$1.  

Revenue = units * price.  With search you have very high prices (RPM) and a large number of units (queries).  With professional content you have moderate prices, but limited units as there is a semi-variable cost of unit production (creating content takes writers, editors, directors, actors etc.).  With the “everything else” category, you have very low prices and a massive number of units.  And as Jeremy Liew showed us, the number of units (users, page views, time) required to build an interesting business is astronomical on low RPMs.

There is a huge amount of money going into new advertising technology, as the biggest business issue facing the consumer internet right now is the price variable of the revenue equation.  Unless someone develops something to improve price per unit, the investment dollars available to consumer internet companies will eventually shrink dramatically.

The Do Not Call Registry does not = the Do Not Track List.

But, you may argue, “who cares about entrepreneurs, investors, and advertisers?”  What about consumers?  The Do Not Call Registry will harm consumers in several ways:

1.  If companies cannot increase yields per unit through advertising, they will look for alternatives.  At best, free content paid for by ads will be replaced be freemium models.  At worst, ad powered innovation will slow down dramatically.  In addition to decreasing innovation and increasing costs, more features and products driven by subscriptions will lead to less openness.  Did you see that story in the Wall Street Journal online today?  If you don’t pay, no content for you.

2.  The line between targeted advertising and merchandising isn’t black and white.  Are Netflix and  Amazon recommendations behavioral targeting?  Amazon clearly states (see below) that these recommendations are based on “items you own” (past purchase behavior) and “more” (your search and surfing history, perhaps).  Aren’t they targeting a product to me for the purposes of making a sale based on previous online activity?  This feature has helped me to discover many great books — is the Federal government going to take that away from me for my benefit?

3.  What about behavioral targeting for content?  The promise of personalization is awesome.  If that promise requires leveraging past data to deliver content to the user, is that okay?  If so, why is it okay to target content based on BT but not advertising?  

These are the types of questions that I would prefer not to have our government answer for me.

Game theoretic implications.

As if having the Federal government consider this type of thing wasn’t cause for concern enough, the firms that you would expect to take the other side of the argument may not have the incentives or current stature to do so.  

Today Google appears to make minimal use of behavioral targeting outside of Doubleclick.  With a significant lead in search, they can always leverage a user’s most recent queries to target ads in and outside of search (e.g., display ads on a content partner’s site based on user’s searches today).  Even if that strategy doesn’t work, with the VAST majority of their revenue coming from search and major competitors powered by display, it may make sense for Google to take a little pain in order to weaken current and future competitors.  Google would benefit from no BT, as long as BT excludes search queries from a user’s current session (with a generous definition of session — let’s say 24 hours).

Yahoo!, Time Warner AOL, and Microsoft are the biggest existing beneficiaries of BT today.  But Microsoft has its software business to power cash flow growth and can weaken its competitors (with the possible exception of Google) by letting this happen.  Microsoft clearly has the incentive to have BT broadly defined to include in-session BT advertising — can you imagine Google without those ads on the right-hand side of the search results page?

That leaves us with Yahoo! and AOL, both weakened players.  Both are fighting for their survival right now, and Yahoo! is likely still feeling pain in Washington from the China writer jailing episode and must be thinking about how best to protect their ad deal with Google.

Conclusion.

Protecting consumers by limiting the way companies share personally identifiable information like email address, name, address, and phone number makes sense to me.  Thinking about how we can protect children online makes sense to me.  Asking sites to be transparent about their policies makes sense to me.  But pressuring companies, under the threat of regulation, to take steps to limit targeting is bad policy and will ultimately harm the group the government claims they are helping — the consumer.