Bryan Semple begs to differ. Semple is a vice president at FairMarket, which sets up on-line auctions that corporate customers use to sell goods. He says it’s too soon to tell just how e-commerce sales will behave. He says the Internet allows for a host of approaches that aren’t possible in the physical world. These approaches—such as more personalization of products and services, or more flexible pricing—could keep customers from following their normal path, Semple argues. Will customers exhibit the same old tendencies? If they do, how can you best prepare yourself? Read on.

PETER FADER: I’ve been investigating grocery-store scanner data for 13 years, watching people buy soap, toothpaste, and breakfast cereals. Now, I’ve taken models I’d developed for these kinds of products and applied them to new types of data such as visits to Amazon.com or CD purchases on the Internet. I’m taking data from third-party sources like Media Metrix and from specific companies and fitting the data into the models that I and others have been using for decades. I’ve applied this approach to the dozen or so leading sites that make up a particular market, such as books, CDs, and travel services.

It’s amazing. The models work better on-line than they do tracking and predicting behavior off-line.

I’ve come to the simple conclusion that the Internet is just another distribution channel, that there’s nothing new under the sun when it comes to looking at people’s shopping behavior.

There’s no question that the process that’s leading people to make purchase decisions is changing. But if we take the 30,000-foot view, where people are running around like ants, and we watch customers choosing Web sites, it’s no different than watching them choose brands of toothpaste or detergent off the shelf.

It has been well-documented in the packaged-goods domain that customers initially get excited about something, but that their interest and purchases then level off and eventually decline. The assumption has been that the Internet is different because it’s an exciting, evolving area. But, when you look at actual sales patterns, it’s quite the same.

To be honest, I have not looked at auctions and other nontraditional ways of buying things, but there’s no reason to believe that, at the individual level, behavior would be any different. It’s your job to tell me that it is.

BRYAN SEMPLE: I agree that on-line buying patterns are similar to what you’d see in a traditional retail environment. But, clearly, you’re not saying that the volume of e-commerce is going to decline.

FADER: I’m not saying that at all. Sometimes people get the wrong impression.

We see all these steep hockey stick curves representing Internet activity because there’s a constant influx of new people to sites. Once some kids’ drink product has been on the shelves for a month, new people don’t enter the market for that drink. But a Web site like CDNow has all these new people coming into the market as Internet usage grows.

What I do is watch a fixed group of people who visit a site, something that most companies don’t do very often. When you look at things my way, you see that, on an individual basis, purchases very quickly flatten out or decline, just as you’d see if you tracked a packaged goods product.

This influx of new customers to Web sites will continue and will mask the erosion that comes as existing customers slow their purchases. At some point—I’m actually starting to see it now—the influx itself will slow, because there are only so many more households, at least in America, that haven’t yet gotten on-line.

SEMPLE: I’m actually not surprised by your results. Most e-commerce is still pretty primitive. We put up a site, market it, go do a large ad deal somewhere, drive traffic to it, and sell people stuff. We’ve basically just taken the mail-order model and made it electronic.

But as technology and techniques improve, I’m not sure we are going to continue to mimic what happens in packaged goods.

Commerce models will fundamentally change. We’ll see, for example, more dynamic pricing [changing prices moment by moment, based on criteria such as supply or the buyer’s profile]. The research we’ve seen coming out of Forrester Research on dynamic pricing shows that it encourages repeat purchases.

We’ll also see "network commerce" become more important. CDNow is running a cross promotion with the Gap. You buy Gap khakis, and you can get CDs. The whole concept of going to a stand-alone site to do something will start to go away.

I think there’s a whole realm of additional network-based business models that may invalidate your data going forward.

FADER: The idea of getting lots of vendors to interact with each other is really no different than getting a bunch of different vendors under the roof of a department store such as Macy’s. And the way that Macy’s will get customers to cross buy is no different than what CDNow and the Gap might do. The patterns we’ll see, as far as people’s repeat visits and purchases, will be no different, either.

SEMPLE: We need to be careful. Right now, Macy’s doesn’t know me. Macys.com, on the other hand, will eventually get to know me and generate a stronger relationship. Now, when I walk into a large department store, it serves up all kinds of brand images I have no desire to see. If I’m cruising through macys.com, it will eventually know that I’m a Polo buyer and will focus my attention on Polo products. The personalization and targeting that are possible with e-commerce will start to make a significant impact. Those processes are still in their infancy.

FADER: I think the benefits and impact of personalization and customization are way overstated. Companies say technology will let them form relationships with individual customers. I don’t see that happening. I don’t think customers want to have personal relationships with all the different vendors they occasionally interact with. Besides, the Internet makes it easier to find the same item at various sites and compare prices, a fact that can reduce loyalty.

SEMPLE: I would argue that, five years from now, price won’t be an issue. It will be whatever the market price is at that moment, regardless of where you buy something. With fair-market pricing, we’ll see a lot of the customer churn go away. It’s there now because there’s a lot of money out there funding dot-com companies. The number of offers getting thrown at consumers to go use a service is out of control. At some point, we’re going to develop a more rational customer-acquisition strategy, and pricing will become less of a factor. How would customers in your model pick where they wish to purchase then?

FADER: You might not want to hear this, but, at the moment, pricing doesn’t matter nearly as much as people think it does. If it did, we would be seeing much broader product searches. For any given category of product, the average number of Web sites that a person will actively visit in a month is rarely more than 1.5. Sure, there are people who visit plenty of sites, but, for the vast majority, price doesn’t matter as much as some notion of convenience or perhaps the appearance of the site or the feeling of being treated well.

SEMPLE: Considering Amazon’s market share in the book industry, it has the potential to skew the data. And remember, people purchase from Amazon because they’re getting the best price.

FADER: But they’re not. Amazon is actually charging a premium compared with other sites. All sorts of other sites are undercutting Amazon on a regular basis. But Amazon can get away with higher prices. If we look at other commodity categories, where there is more market-share parity, price should be paramount, and we should see much broader searches. We don’t.

SEMPLE: I’ll buy that. If you can provide other benefits, you can charge a modest premium. How much of a premium are these guys charging over the other channels?

FADER: Apparently not enough. Perhaps that’s why Amazon is still losing money. I really believe that. Amazon should jack prices up and emphasize the convenience, selection, and service.

SEMPLE: But fair-market pricing is certainly going to limit the range of prices a retailer can charge because it is simply too easy for an on-line customer to find another retailer. I may be willing to pay a 10% premium to have superior customer service, a great interaction with a site, and great shipping, but perhaps not 12%.

FADER: Still, companies are making a big mistake if they assume pricing will be a neutral enough issue that they can focus exclusively on building a community of customers. There are so many competing communities, and consumers are inherently polygamists. They don’t want to form relationships as much as the firms want them to or think they want to.

Right now, there appears to be a lot of monogamy going on because consumers aren’t yet educated enough about their options. But I see explosive growth in shopping bots [software that surveys numerous Web sites so a customer knows where he can get the best price].

SEMPLE: That was my point. Shopping bots help people get the lowest prices on products. In addition, some of the top portals—MSN, Excite, ZDNet—help people research products, get pricing, and then purchase. Over time, prices will become common enough that customers’ buying decisions will move on to other issues. Price is just too easily measured, displayed, and compared against. It sounds like what we fundamentally disagree on is the role that new forces on the Internet will play. You don’t think that any of the new types of on-line shopping experiences—dynamic pricing or personalization—will fundamentally change the model of behavior?

FADER: I don’t want to say those approaches are going to fail. I simply think some will work for some firms for some of the customers they serve. There’s no question that some customers will like the idea of getting pants customized to their size. Those people will flock to the site that lets them do that, and it will get a larger share. But when we look at the distribution of market shares across the various players and the switching patterns of the consumers, those will be no different than they are now.

SEMPLE: Then what will drive buying behavior in the future?

FADER: You’re exactly right in saying that technology will cause shifts to occur. If you can come up with clever pricing mechanisms or e-mail relationship management, you will drive more people to your site than others do to theirs.

What scares me is too many companies say that, because there are new technologies and new ways of driving people to a site and keeping them there, the old models and the old statistics don’t apply anymore. That’s exactly where I take issue. It’s very important to marry the old ways of looking at data patterns and the new ways of bringing people into a site to understand what’s happening. The old models will give you a very good sense of whether your new pricing mechanism or loyalty program is working.

Also, too many companies currently fall into the trap of saying, We are making the whole category grow. We’re going to involve consumers and find out what they want and tell them about it before they even know the products are available. We’ll make people feel so warm and fuzzy that they’re going to buy more CDs or books than they would have before.

SEMPLE: The technology does let companies do some important new things. For instance, we’re solving a classic problem for large companies. They have surplus, off-lease, or remanufactured items. The on-line auctions that we organize for them are giving them a completely new distribution channel in which to move those goods. And the companies that have these auctions on their sites are finding that their sites end up being incredibly "sticky." There’s an amazing amount of repeat business from consumers.

FADER: That’s great. The only problem is that the process is so new, you don’t have benchmarks to tell you whether the level of repeat buying is higher or lower than you should expect. There are well-known ways of understanding what the expected level of repeat purchasing ought to be. You see someone has bought twice over the past six months. How many times should you expect them to buy over the next six months? The naive answer is to say two. It doesn’t usually work that way. Some people speed up. Others slow down. You need formal mathematical models to give you a sense of how much speeding up or slowing down will occur. I see so few firms setting benchmarks because they think that the Internet is a whole new world.

SEMPLE: We track the amount of repeat purchases that we get. But it’s possible we’re not going to see a consumer who buys a computer coming off a lease for another two years, and it makes the process challenging.

FADER: Because you’re dealing generally with bigger-ticket items than breakfast cereals, the honeymoon period before purchases fall off is going to occur over a much longer period. Still, when we look at this five years from now, we will see that same pattern occurring. There are things you could do with your data now to anticipate how, when, and if the slowdown will occur. Don’t think that a dropoff is impossible because this is a new world.

SEMPLE: My belief is that these are crazy times right now in the on-line world. E-commerce is a very long game, and we’re just starting it. There is currently a lot of noise: the newness factor, new users coming on board, and massive amounts of money that are getting thrown at promotions. I think that any great conclusions drawn from the current phase may not hold after another year or two.

FADER: The actual underlying behavioral patterns—honeymoon, and then slowdown—will hold. The issue is: When will companies finally come around to seeing them?

SEMPLE: You honestly don’t believe that personalization or the other things that happen in the on-line world will overcome the tendency for a honeymoon period, then a slowdown?

FADER: The phenomenon might stretch out, becoming slower or less dramatic, but it will occur. And companies need to start thinking about how they’ll react to it.

Right now, I find two camps of companies. Some are worrying only about the big-picture business issues. The market is growing so rapidly that they say, Let’s just put products out there; we’re making money, so let’s not worry about details yet. On the other extreme are companies caught up in what I call metric minutiae. They’re trying to analyze the data they have on customer behavior but are caught up in the details and don’t know how to make sense of it.

A huge gap exists between these two philosophies. Companies need to do analysis, but only in manageable chunks. I don’t see companies in that happy middle ground.

SEMPLE: Are you saying that companies that are in the happy middle ground can avert the honeymoon period and the inevitable slowdown?

FADER: They can soften the drop a bit, or they can at least anticipate that it’s coming. They can know that the people who are buying tons of stuff now are going to slow down. They can then reallocate resources to try and pull in more new people rather than bribe existing people to stay around, when those people don’t really want to.

Understanding the patterns that occur leaves lots of possible managerial action. Anticipating the patterns and asking the kinds of questions you’re asking is exactly where industry needs to go, and it’s going to get there. Attitudes will change when the big shakeout occurs. I don’t know when that’s going to be.

In the end, being able to sort people, from the first couple of transactions, into bins of good and bad customers will buy you much more than trying to understand the needs of each individual customer.

SEMPLE: You’re saying, though, that if people use analysis they can soften and change the up and down behavior?

FADER: Yes, they can. They’ll still have the honeymoon period and the slowdown. But they could change the nature of it.

SEMPLE: So there is hope.

FADER: Absolutely. Companies just have to start asking the right questions and thinking about how to address the right issues.

Fader can be reached at fader@wharton.upenn.edu. Semple can be reached at bryan.semple@fairmarket.com.