September 2008 - GoogleNomics
Hal Varian, the online giant’s chief economist, explains some of the economic forces
and corporate strategies that drive Google’s business -and that of e-commerce as a whole.
By Tom Dellner
When you’re Google and you’re looking for a keen economic mind to help you refine your auction model, do some forecasting and statistical analysis, you don’t hire the top-of-the-class graduate with a freshly minted master’s or Ph.D. from a world-class university. No, you hire the person who taught that top-of-the-class graduate.
In other words, you hire Hal Varian. Varian is an MIT grad who has a master’s in mathematics and a Ph.D. in economics from the University of California, Berkeley. He has taught at Berkeley, MIT, Stanford, Oxford, Michigan and other universities around the world. He’s written textbooks that have been translated into 22 languages in addition to a regular column for The New York Times. He’s also co-authored the best-selling book on business strategy, “Information Rules: A Strategic Guide to the Network Economy.”
Electronic Retailer sat down with the personable Varian for a fascinating discussion that covered everything from the origin of the online auction model to the effect of the traditional Spanish siesta on e-commerce.
Electronic Retailer: “Chief Economist” is a title you don’t encounter every day. Could you explain your role at Google and how it first developed?
Hal Varian: I first began consulting with Google six years ago. When I arrived, Eric Schmidt [Google's CEO] said to me, “Why don’t you take a look at this auction model? I think it might make us some money.” So I started looking at the ad auction. It had just been put in place. I did some game theoretic analysis on how the auction worked and it turned out to be pretty useful in future development. Then I turned from that to do some forecasting, because in those days, the way we did it at Google was to take a column of numbers in Excel and hit the forecast button; I suggested there might be a better way. So I worked quite a bit on our econometric forecasting for queries and revenue.
Now we have a group of about 40 people (statisticians, econometricians and some operations research people) and we do a lot of the high-powered statistical analysis that drives many of the business processes at Google.
ER: After a career in academia, what has it been like to apply your skills in the corporate world?
Varian: If you’re an academic, you’ve got months or even years to work on a project. Then you publish it and it turns out that nobody cares. At Google, you’ve got weeks or maybe days to work on a project, but once you reach a conclusion, it could involve millions of dollars for the company. So it’s a much tighter time scale, but the outcome has a much greater impact.
ER: You’ve spoken of Google as being “recession resistant.” Could you elaborate?
Varian: The reason we’re recession resistant, in my opinion, is that search-based advertising (our primary business, at the moment) has very measurable results for advertisers: you spend a certain amount of money and receive a certain amount of clicks or conversions. So our business is very tightly integrated with the sales process itself. In a sense, it’s the cost of making a sale. And the last thing businesses want to cut during a recession is something that’s an integral part of making sales.
ER: It’s also been said that e-commerce as a whole is well-positioned to weather an economic downturn. Do you agree?
Varian: I do. I think it’s still perceived to be the case that shopping online gives you better prices and more selection. So when times get a little bit tough, people might shift some of their spend toward online businesses. This is a little bit speculative, but I think it holds true. Increased gas prices might factor in a bit, as well, but my personal opinion is that this is a smaller factor than some are making it out to be.
ER: What are some of the ways that the economics of online commerce diverges from that of traditional commerce?
Varian: The one way that I think is most interesting-and perhaps the least explored-is the social aspect of shopping. When you look at people who go shopping, they tend to do it in groups, and we haven’t yet found an effective way to reproduce that experience online. It’s still something of a solitary experience; it doesn’t have that same social component as going to the mall with your friends. We’re beginning to see some aspects of this with the continued development of social web applications, so maybe we’ll see some movement in this area.
ER: You have written about the Japanese concept of Kaizen. Could you please define the term and explain its application to online commerce?
Varian: Kaizen is an old term that means “a process of continuous improvement.” It was frequently applied to the Japanese method of manufacturing from the ’80s that had all of the American auto companies concerned with the productivity advances seen with Toyota, Honda and others. It was also applied to other production processes-in addition to automotive assembly-such as consumer electronics manufacturing. But I think it has a particularly important application to e-commerce. For example, at Google we have hundreds of experiments running at any one time. We’re always trying to improve the way our search-function algorithm or ad-ranking algorithm functions, or we’re seeking improvements in user interface. The idea that you can continually experiment-through multivariate testing, for example-and improve your product is analogous to these production process improvements from the ’80s. But online, it can be done with a speed and efficiency that just isn’t possible with most types of products.
Perhaps the lesson that should be taken from this is that when you design your system to do whatever it is you want it to do, you need to set it up so that you can easily do experimentation with it. This requires quite a bit of discipline-lots of times when you’re designing a new system, the biggest goal is to get it up and running and out the door-but it’s worth a few extra weeks to make sure that it can be tested, tweaked and evolved. Many companies tend to forget that in the rush to get a product to market.
ER: The Internet makes it extremely easy to research product benefits and seek the opinions of others. Also, with comparison shopping tools, one can easily find the best price. Do you think this commoditizes goods that heretofore were anything but commodities? What are the implications and how should retailers respond?
Varian: In a sense, this is just the continuation of a process that has been going on for literally hundreds of years. For example, it used to be that every pot and pan was handmade. Then we went to mass production. Then we started selling wholesale to stores. Now people go online where it’s very easy to compare products and read reviews and find the best pot and pan for their needs. So it’s not a discrete break from what we’ve seen for centuries; it’s really the next step in the evolution of a more efficient market. So I think we’ll see the same sorts of forces at work: vendors will try to produce the highest-quality products at affordable prices and market the benefits in a way that appeals to different market segments. It’s just that now we do it over the course of months, in some cases, instead of the years or decades it took in the past. The evolution has been sped up by the efficiency of the market.
ER: Google’s auction model, with its quality score that rewards the best-performing ads, has been the subject of much debate. Many see it as an incentive for advertisers to create better offers and ads, which then benefits consumers and, in turn, Google. But others complain about a lack of transparency-that Google operates as a “black box,” deciding who can and cannot advertise on its system. How would you respond to this complaint?
Varian: We’ve certainly heard this complaint from advertisers, and we’ve listened and we understand the issue. One of the things we’re developing-which we’ll be rolling out in the next several months-is a set of tools that we think will help advertisers bid more effectively. Our big push this year is for better effectiveness measures, better reporting and better tools for advertisers. In fact, just last week we came out with the search insights tool that helps advertisers better understand factors like seasonality and other underlying trends that affect their ads’ performance.
Also, I should remind you that the model Google uses is the same basic model that everyone in the industry uses-Microsoft and Yahoo are doing essentially the same things. And being able to estimate ad quality in an accurate way is something that’s very important to success in this industry. The challenge is to communicate more effectively with the advertisers so that they can bid more intelligently.
ER: I understand that Google’s data reveals interesting trends about how world events and other factors affect your business and, ultimately, e-commerce as a whole. Could you give some examples?
Varian: We certainly saw the impact of the European Cup-people were more interested in watching soccer than clicking on ads-and we’ll definitely track the effect of the Olympics. Moderate bad weather seems to create an uptick, although severe bad weather-and, obviously, power outages-will cause a noticeable downturn. We see all this in the data. Even the daily tempo of life in certain areas is reflected in the data. For example, we can see the impact of the siesta in Spain with a little dip in the query traffic after the relatively late lunch they have. There are scores of examples like that where you can gain insight into human behavior by looking at the active data
ER: The lack of cross-border e-commerce is somewhat surprising. Do you think we’ll see growth in this area?
Varian: I think we will see some growth. Again, there are cultural factors at work-different countries have different practices with respect to credit card usage, the reliability of the mail systems or just how comfortable people are shopping remotely. It takes time for people to become comfortable with remote shopping, but we are seeing strong growth in online shopping and advertising globally, so I think we’ll get there.
ER: In your opinion, which developments in e-commerce have come as something of a surprise to economics experts?
Varian: I think, overall, people have been a bit surprised at how common the auction model has become online, although there were intimations that it might work well. The second thing is the rise of advertising as a business model. Advertising started quite early on the Internet, but it was all impression-based. It was the model that people were familiar with, but it saw relatively slow growth. However, once we turned to click-based advertising, which is essentially a direct marketing model, we saw extremely rapid growth. So it wasn’t surprising, I suppose, that advertising would play an important role, but what was not understood was the critical difference between traditional display advertising and response-based advertising.
ER: Do you foresee significant changes in online commerce over the next few years?
Varian: I think the big phenomenon will be the mobile platform with Internet access, which will allow for all sorts of new commerce models. I think we’ll also see the integration of payment systems with mobile devices. And I expect to see some evolution in social shopping-maybe it will take a mobile context or perhaps a desktop context; that remains to be seen.
Marketing Services, Online-Internet | 9 Mar 09