Arnold Kling  

Hal Varian on Googlenomics

PRINT
Movie Review... Superheroic Dominance in U.S. ...

the Wall Street Journal does an interview.


WSJ: Is Google an example of a “winner take all” market, where “network effects” make it more valuable as it gets more users?

Varian: I don’t think [search] is a winner take all market at all, because it’s so easy to switch. You type a different URL into your browser. For eBay, the buyers want to go where there are the most sellers are and the sellers want to go where the buyers are. [In search] the advertisers want to be where the most users are [but] the users don’t want to be where there are the most advertisers.


Here is an interesting notion:

Varian: I think marketing is the new finance. In the 1960s and 1970s [we] got interesting data, and a lot of analytic fire power focused on that data; Bob Merton and Fischer Black, the whole team of people that developed modern finance. So we saw huge gains in understanding performance in the finance industry. I think marketing is in the same place: now we’re getting a lot of really good data, we have tools, we have methods, we have smart people working on it. So my view is the quants are going to move from Wall Street to Madison Avenue.

One of the things that I like about Hal Varian is that he has never been an academic snob. Many years ago, for example, he pointed me to The Economy of Ideas, an essay by John Perry Barlow, who had little in the way of professional economic credentials but a lot of useful insights into the issue of intellectual property in the digital age.

I'm not sure that I agree with Varian about the parallels between financial data and marketing data. In finance, the questions are fairly obvious: how can you predict returns? how do you estimate variance, and what do you do about it? etc. So you know what you want to do with the data--the trick is discovering relationships that continue to hold out of sample. People with better technical knowhow will have an advantage.

I would guess that the issues in online commerce are somewhat more open-ended. It may be relatively easy to get reliable answers using even crude statistical tools. The challenge is to come up with the most useful questions.


Comments and Sharing


CATEGORIES: Business Economics



COMMENTS (7 to date)
Biomed Tim writes:

"I'm not sure that I agree with Varian about the parallels between financial data and marketing data. In finance, the questions are fairly obvious..."

How are the questions for marketing less obvious?

Steve Sailer writes:

I worked for the first company to use supermarket scanner data in marketing research, and for its one competitor. The marketing data industry has a tendency toward winner-take-all monopoly (e.g., the Nielsen TV ratings), but in the scanner data business, the DOJ prevented the two leaders from merging in 1987 and the pugnacity of the two competitors caused a subsequent price war that went on for over a decade, with disastrous effects on the stock price.

academicSnob writes:

It was Hal Varian who pointed me to aimst - most certainly he is not an academic snob.

Amol Shelat writes:

"I'm not sure that I agree with Varian about the parallels between financial data and marketing data...I would guess that the issues in online commerce are somewhat more open-ended. It may be relatively easy to get reliable answers using even crude statistical tools. The challenge is to come up with the most useful questions."

Actually, the questions in online commerce and in marketing are not as open-ended as you might think. Given the level of competition in the consumer products industry for the consumer's attention, the questions that marketing research companies receive have become both incredibly pointed and diverse. In response to this, the level of mathematical and statistical sophistication in marketing models has greatly expanded. To see this, you have only to look at the people that these firms nowadays hire: statisticians, computer scientists, mathematicians, and social scientists (usually from among some of the top universities around the world). If I'm not mistaken, these are same pools of people that major banks focus on in order to fill their ranks of bankers and quants.

Here's a scenario of a typical question posed to marketing researchers today. Suppose a company designed and implemented an Internet media campaign. Now, they desire to know exactly how much incremental volume that particular Internet media campaign generated in Chicago. In order to answer questions so-focused as this, marketing research companies must rely upon a dazzling array of complex and rigorous statistical models. This is not your ordinary, run-of-the-mill regression stuff, either. In fact, you would be astounded both with the predictive accuracy and level of detail that these models now produce!

I should note, though, that I work in the marketing research industry.

Steve Sailer writes:

There's nothing about Internet marketing data that's radically more complicated than the supermarket scanner data available in the 1980s. The marketing research business was employing sophisticated quantitative analysts back then -- heck, they had them in the 1970s before there was enough data.

The key point is, though: people in marketing don't get paid as well as people in Wall St.

Varian is definitely not an intellectual or academic snob. He answers his e-mail, for one thing. For another, in 2002, he and I ended up debating each other over some things he'd written in 'Information Rules' about the work of a couple of friends of mine.

He lost the debate (David Friedman thinks so too), and resisted conceding so, but he was very nice throughout.

Matt writes:

I always hear that businesses should solve problems. So, I am told, ask the customer what are his biggest problems. Narrow the field, but ask for the problem.

Customers, if they knew the solution, would not have the problem.

Comments for this entry have been closed
Return to top