more efficiency means less profit


I’ve been following the “Do Not Track” discussion with interest. I’m not sure I have any particularly strong feelings about the proposed implementations, but I’m curious to see what happens. And of course it’s been hilarious, as always, to see some of the knee-jerk pro-corporate responses that the policy proposal has prompted.

There’s an aspect of the debate that I briefly tweeted at Harlan Yu, and I guess it might be worth expanding on a bit. Consider an excerpt from this post over at TLF:

The FTC’s report calls for a “uniform and comprehensive” way for consumers to decide whether they want their activities tracked. The Commission points to a Do Not Track system consisting of browser settings that would be respected by web tracking services. A user could select one setting in Firefox, for example, to opt out of all tracking online. The FTC wrongly calls this “universal choice.”

Really, it’s a universal response. It’s a single response to an overly-simplified set of choices we encounter on the web. This single response means that tracking for the purpose of tailored advertising is either “on” or “off.” There is no middle setting. But it is the “middle” where we want consumers to be. The middle setting would represent an educated setting where consumers understand the tradeoffs of interest-based advertising – in return for tracking your preferences and using them to target ads to you, you get free content/services. But an on/off switch is too blunt and not, err, targeted enough. There is no incentive for consumers to learn about the positives, they’ll only fear the worst-case scenarios and will opt-out. In return they’ll also opt-out of the benefits. [more on the “middle” below].

(Aside: can we please stop with the transparent attempts to frame the debate in a self-serving way through use of the word “nuclear”? Everyone knows what you’re doing, dude.)

Let me quibble with the free content/services bit embedded here. Implicit in this assertion is the idea that advertisers’ ability to possess better information about consumer behavior allows ads to be more precisely targeted, making the ads more valuable to the businesses that place them. This increases the revenue available to ad-supported industries, many of which provide valuable content or services through what is a de facto tax on businesses.

I think the evidence for this dynamic is weaker than a lot of people suspect. As far as I can tell, it’s all based on Google. GOOG showed up and provided contextualized ads to consumers and a model that allowed advertisers to only pay for purchases that were “working”. This is pretty much the only way they make money, and they make a lot of it.

But Google’s a huge success in a landscape of failure. Online ads sell for pathetic rates relative to broadcast or print. This is because by all accounts online advertising doesn’t work very well. You can measure whether someone clicks on an ad, and often whether they buy something after that click. But it turns out they rarely do those things. So businesses aren’t willing to pay very much for ad space on websites.

Is it really a coincidence that the advertising medium with the best instrumentation also appears to be the least effective? I suspect it’s not. It may be that ads never worked as well as the industry had told us; or it may be that the eyeballs/clicks/conversions funnel is a naive conceptualization of how the system works. Either way, Google has succeeded by giving advertisers what they think they want, which is analytic tools that seem to reveal that the whole enterprise is horribly ineffective.

I think the push for better tools and more efficient ads is basically a race to the bottom. In fact, less perfect instrumentation might allow the ad industry to capture a bit more revenue from business thanks to decreased efficiency. If I’m right, those in the business of selling ads should be excited about initiatives like Do Not Track. I don’t think it’s conceivable that the market will cease pursuing greater efficiencies unless this kind of regulatory intervention occurs.

It might sound a little strange to hear me sign off on the government propping up an industry I detest. Honestly, advertising is a big reason I find myself on the liberal side of the corporate/state distrust spectrum. There aren’t many things I find more odious than cynical attempts to manipulate people’s thoughts and desires — especially when it’s done using quirks of biology and psychology. Our anti-propaganda norms make it quite clear that we don’t tolerate such manipulation from government; when examples of it are revealed, they’re harshly criticized and punished. Yet we accept as perfectly normal the idea of such manipulations being pursued by private entities, who are free to try to cajole us into doing things contrary to our own interests, and without any mechanisms for controlling them other than the adeliberate market and (viciously resisted) regulation.

It’s a bit weird. Distressingly weird! So I think preserving personal privacy is probably worth the introduction of this kind of inefficiency. And I think the preservation of the media creation business is desirable, and I don’t really believe in any models for accomplishing this except some sort of (probably convoluted) public subsidy. If I’m right, decreasing the efficiency of advertising will act as a de facto tax on businesses which choose to advertise. We’ll consequently pay slightly more for products sold by such businesses, and the difference will help pay for our magazines and newspapers and web videos and music. I think that’d be an okay deal.

About the author

Tom Lee


  • My guess is that the gap between online and traditional ads largely reflects various kinds of inertia. There are lots of large, bureaucratic companies who have been buying print and television ads for decades. Large, bureaucratic companies change their behavior slowly. Therefore, as eyeballs shift from old media to new, those old institutions are going to continue buying a dwindling stock of “old” eyeballs while people struggle to find buyers for the growing stock of “new” eyeballs. But eventually, those old eyeballs are going to go away entirely, at which point the same institutions are going to have little choice but to buy online ads. In any event, I don’t think the current differential tells us very much about what the long-run equilibrium will be.

    On advertising as manipulation: one of the most important trends in online advertising has been a shift from “brand” advertising of the sort that does seem manipulative to “informational” advertising that informs customers of the existence of products that the advertiser has some evidence they actually wants. Google’s keyword advertising is a sterling example of this: you only get ads for cars if you do a search on a car-related search term. And this is, not surprisingly, the most profitable kind of advertising. Arguably, companies rely on the manipulative sort of advertising the most when they know nothing about their audience and are therefor reduced to using psychological tricks to get the results they want.

  • Although I think that Tim right that the long term equilibrium is hard to predict, I think that Tom is right that some sort of subsidy is going to be required over the long run. There’s too much value being produced by people making non-rival informational goods (and too much potential for more that is being squandered because no one can afford to do it even part time). I suspect that it’s going to take some new economic thinking, though, since the ‘pretend it’s a rival good and force a market via legislation’ model that we’re currently working in isn’t really serving the interests of anyone who isn’t big enough to make the legislation part happen.

  • Tim: maybe so. I’m less sanguine about online ads’ potential. There’s still an order of magnitude standing between print and digital advertising sales. I agree that closing this gap is conceivable, but I think we’re past the early days where optimism made sense as a default.

    In terms of informational vs. persuasive ads: I’m reading The Master Switch now, and I know you’ve finished it. Wu makes the point that the early days of radio advertising involved just such a model — persuasion was only introduced gradually, and became more insistent over time. If informational advertising is really more effective, why would that be (perhaps Wu discusses this; if so, I haven’t gotten there yet)? I suppose the best answer at hand might be “personalization.” I’m a little skeptical of that: I suspect the type of material being broadcast goes a very long way to defining the audience and the products they’re interested in. Google’s made incremental improvements to this approach, but is it really a complete break from the past?

  • Yet online ad spending accelerates. I an millions of others watch Youtube, and we can’t ignore that stupid Oprah OWN ad.

    I don’t really see any quantitative analysis in this post. Finer measurement of data is a good thing, companies save on spending, Google has results it can boast about, viewers get more relevant advertising. In contrast to the past where small businesses may be hesitant to engage in print advertising in random coupon books, they are now in more control, and therefore, more willing to spend and advertise.

  • Here’s some actual numbers for traditional media.

    Online = $5-30 CPM
    Network/Local TV = $20 CPM
    Magazine = $10-30 CPM
    Newspaper = $30-35 CPM

    Hulu is also doing better than traditional TV. Running an ad during The Simpsons on Hulu, for example, costs about $60/CPM, Bloomberg reports; running the same ad during prime-time on TV costs about $20-$40/CPM—or over 60 percent less in some cases.

    Your premise that “Online ads sell for pathetic rates relative to broadcast or print. This is because by all accounts online advertising doesn’t work very well.” seems rather dubious.

  • My intent was to discuss a structural problem, not to perform a quantitative analysis. The back-of-the-envelope figuring you link to leaves me unimpressed, I’m afraid. It’s easy to find similar accounts that find exactly the opposite result (or even analyses that make it look like web attracts higher CPMs than print). All of these suffer from accounting tricks and apples-to-oranges problems.

    The one figure that everyone seems to agree on is that the online market is about 10% of the print market. It’s not surprising that growth rates are high — that’s what tends to happen when your base is small. But you can’t count on that effect lasting forever. The concern is that even for publications where the audience is well-developed for both print and online (e.g. the NYT) there remains a large disparity in what can be charged for the online medium. And of course the situation is even worse for smaller brands.

  • Your 1st link is about the NYT’s paper revenue vs its online revenue. It’s not surprising that a newspaper would sell more on paper. The publishing2 also sloppily calculated CPM based upon face figures rather than the undisclosed actual numbers. I go to nytimes every now and then because of memeorandum or other sites, but I don’t read it thoroughly like I would its paper, that certainly contributes to the disparity. The new york times isn’t exactly thrivingnyin the age of the web anyway (stock at its 1984 levels), not sure why you would study them.

    I don’t see the structural problem that you are describing. Google Analytics, Quantcast, other data analyzers have existed for a while. Because of them, companies are better able to target certain demographics. There is limited ad space for Archer on Hulu. Knowing the behavior of the demographics and the ads response, companies are more willing to bid up the cost per impression for the space. Contrast that to the newspaper biz, which with just a 5% dip in <a href=""circulation, flatin 2009, online ad revenues continued to climb to $24.2 billion, while print ads fell 28.6% to $24.8 billion. Online ads are expected to rake in $34.4 billion by 2014, which means print ads should dip below their online counterparts in a matter of months.

  • What if better-targeted ads aren’t more valuable to the advertisers? Tracking makes it easier for advertisers to conceal their level of spending, and signaling the willingness to back up a product with “wasteful” ad spending is the main economic reason for advertising. If it’s hard for a potential customer to tell a mass buy from a targeted buy, the signal isn’t being sent. So maybe a privacy-hawk Web would command higher ad rates.

    Web ad targeting: can customers get a better deal?

    Ad targeting: better is worse?

    (The second has some citations to interesting papers on advertising and signaling.)

    (Of course, it would be better to see the problem resolved with privacy technology on the browser side than through top-down regulation, which the major advertisers would pack with loopholes anyway.)

  • I think Don Martin is very largely right here.

    A very intelligent British adman makes the distinction between ads which create sales and ads which create saleability. The kind of adspend-as-signalling Don refers to is very much about creating saleability rather than sales. Conventional media do a better job of this signalling, which may be complementary to – and not replaceable by – money spent online.

    By contrast, because it is more immediately measurable, online activities are mostly valued for their transactional effects – sales, rather than saleability. The lower prices do not necessarily show that online ads are less valuable to the people who buy them, only that online advertising space has less scarcity. Interestingly behavioural targeting further reduces scarcity, and so may make ads more valuable but less expensive.

    Incidentally Becker and Murphy on ads as complementary goods is worth reading here.

    Rory Sutherland, Vice Chairman, Ogilvy, London.

    PS I am not an economist (but I play one on television).

  • I really wanted to write down a quick message to express gratitude to you for some of the magnificent steps you are giving at this website. My considerable internet investigation has at the end been honored with extremely good concept to go over with my relatives. I would say that many of us site visitors are undoubtedly lucky to be in a very good network with very many wonderful individuals with very beneficial basics. I feel quite happy to have used your entire webpage and look forward to some more entertaining moments reading here. Thank you again for all the details.

By Tom Lee