Although many e-commerce companies collect cost and usage data about their World Wide Web sites, few of them understand in any detail how well such information measures their sites’ performance or how this performance compares with that of competing sites. Still fewer companies know if their sites are becoming more effective over time.

That didn’t matter very much as long as venture capitalists and the capital markets were willing to throw money at dot-coms. But since last spring’s crash, investors have been insisting, if not on profits, at least on objective measures of a site’s success in attracting, converting, and retaining visitors.

We have analyzed 250,000 data points describing the behavior of on-line visitors to the sites of hundreds of companies, employing many business models, during the 18-month period from January 1999 to June 2000.¹ This information has helped us achieve a new level of rigor in evaluating the performance of e-businesses. At least six months before the collapse, the data showed that Internet companies suffered from a kind of fatal attraction: they were successful at luring visitors to their sites but not at getting these visitors to buy or at turning occasional buyers into frequent ones. Indeed, the more visitors the sites drew, the more money they lost.

Yet by the time investors finally lost confidence, the tide had actually begun to turn; in fact, the era of business-to-consumer (B2C) profitability is now around the corner. A small but solid group of leaders has been profitably converting visitors into customers and getting them to return.

Fatal attraction

Most measures of e-performance track variations in traffic—page views, advertising impressions served, unique users, and so on. But the foundation of long-term profitability is lifetime customer value: the revenue customers generate over their lives, less the cost of acquiring, converting, and retaining them (see sidebar "The e-performance scorecard").

We studied consumer behavior during two consecutive periods—the first and second halves of 1999—and results have begun to come in for the first half of 2000. In all, we tracked more than 650 million visitors to eight kinds of Web site—portals and communities such as Yahoo! (see sidebar "Community values"), as well as sites for news and general content, specialized content, general retailing, specialty retailing, retail services, new marketplaces, and financial services. These visitors made 2.7 billion visits to the sites, conducted 27 million transactions, and (in 1999) paid $4 billion to 224 companies based in North America, Europe, and Latin America.

Almost none of the e-companies that lavished large sums on e-marketing campaigns achieved above-average visitor growth

Our first discovery—no surprise today—was that the intense early focus of e-businesses on attracting the public to their sites went pretty well universally unrewarded: almost none of the companies that lavished large sums on e-marketing campaigns achieved above-average visitor growth or conversion rates. Businesses that failed to do so included Boo.com (in the United Kingdom), Pets.com (despite the ubiquity of its TV commercials), and Living.com, which closed after burning through more than $100 million. We found on-line brokerages that spent upward of $50 million a quarter, and as much as $25 million on launch campaigns, when total revenues were unlikely to pass $30 million. The market worked out these truths in April 2000, but the numbers had long been plain to see.

Data from the first half of 1999 suggested that fewer than 4.5 percent of site visits resulted in sales and that fewer than 10 percent of visitors who made initial purchases made later ones. By the second half of 1999, consumer choices had proliferated so much that only 2.5 percent of site visits resulted in sales. An increase in repeat purchasing, to 18 percent, seemed impressive, but the net effect was only to undo the deterioration in the conversion rate. Companies were spending a little less to acquire customers but still losing money on each of them. During the first half of 1999, it cost, on average, more than $1,100 to acquire a customer, who would typically spend a total of $400 on at least two purchases in that period. Similar customers in the second half of the year cost only $800 to acquire, though their spending still didn’t exceed $400.

A number of unpleasant facts lurk behind the numbers. First, the content (rather than commerce) sphere is significantly overpopulated. Second, providing decent coverage and continually creating and updating content require a large and talented staff. Third, the high off-line costs of production, marketing, and distribution have been not so much eliminated as replaced by those of on-line technology, customer acquisition, and brand creation. In the first half of 1999, the average site was losing just under $1 million a month; in the second half the losses had actually increased somewhat, to $1.1 million a month.

By the end of the third quarter of 1999, average advertising revenue per visitor had fallen below 50 cents and was still declining. During the second half of the year, average rates charged to advertisers declined by 3 percent for nonportal players and by 12 percent with the struggling portals included. Even the specialty-content sites, which can generally command higher advertising rates because of their focus, weakened a bit during the first half of 2000. The site of one of the most prominent media companies now fills 60 to 70 percent of its advertising space with its own ads.

Meanwhile, subscription revenues for on-line content remain rare. Of the content aggregators in our sample, fewer than 10 percent were garnering meaningful subscription revenues by the second half of 1999. Countering the culture of free content seems to be impossible: consumers who will pay $4 for a magazine at the checkout counter refuse to pay $4 to view it on-line; nor, in general, will they pay an annual subscription fee. The Wall Street Journal and Consumer Reports may now be the only subscription sites with a substantial base.

Naturally, companies have sought to modify their business models in response to these problems, most often by moving into e-commerce. But only about 20 percent of all content aggregators had been able to abandon the pure advertising model by the end of the second half of 1999. Those that moved into e-tailing generally leaped from the frying pan into the fire. The overall growth of e-tailing revenues is admittedly promising; in 1999, they reached $33 billion in the United States, $3 billion in Europe, and $160 million in Latin America, and by 2005 they are expected to multiply sevenfold, to more than 7 percent of retail sales in both the United States and Europe (see sidebar "Performing in Asia"). But the daily failures in this space reflect the fundamental fact that even very exten- sive spending to build brands and market awareness typically fails to achieve breakeven scale, which may range from several hundred million dollars up to as much as $1 billion.²

Light at the end of the tunnel

The fourth quarter of 1999 was an improvement on the third. Visitor bases increased by 64 percent, conversion rates by 65 percent, and revenue per customer by 28 percent; and 12 percent of e-businesses were profitable. In the first half of 1999, quarter-on-quarter visitor growth was 34 percent; the new-customer conversion rate rose by 5 percent and revenue per customer by 80 percent; and 5 percent of e-businesses were profitable. In the fourth quarter, e-businesses, especially transaction aggregators, also kept customers and extracted value from them more effectively (Exhibit 1).

But this apparently encouraging record wasn’t nearly good enough to make e-businesses profitable overall (Exhibit 2). The performance of the content sector was particularly dismal. A detailed view of the data shows that just a small number of players are garnering a disproportionate share of all revenues: 10 percent of the content aggregators’ sites actually accounted for no less than 80 percent of total advertising revenues, and 15 percent of the transaction sites claimed 85 percent of the sector’s total revenues.

Fixing the vicious cycle

But the success of this small number of Web sites indicates that the fatal attraction isn’t an immutable fact of e-life. The leading companies in our survey achieved visitor conversion rates of 12 percent, customer churn rates below 20 percent, and repeat purchase rates of around 60 percent. These businesses generate nearly three times as much gross income from repeat customers as from one-time buyers.

Since the strategies of e-tailers are well known and for the most part easily replicable, the difference between top- and bottom-quartile performers (Exhibit 3) must result directly from superior underlying operational skills in acquiring, converting, and retaining customers. Moreover, contrary to the widespread view that e-commerce is solely about business experimentation, top-quartile aggregators succeeded by marrying innovation to three tried-and-true principles from the brick-and-mortar world:

1. Match value propositions to segments: Top-quartile businesses focus on core product or service propositions that fit the needs of well-defined consumer segments and have relatively sustainable revenue models—that is, such businesses can generate a positive gross margin without achieving monumental scale or unrealistic conversion levels. The resulting average revenue per customer of these businesses is four times that of the entire sample.

2. Control extensions of product lines and business models: Under pressure to generate new revenue streams, many e-businesses have greatly expanded their product lines. This approach to fixing weak underlying operational models has generally led to highly confused offerings and to a loss of management focus. Fewer than 20 percent of the companies in our sample generated significant noncore revenue streams.

   Best-practice players focus on deepening their current offerings and on adding complementary products or content services oriented toward the same customer segment that was targeted by their initial business model. CNET Networks added a comparison-shopping service to its computer and consumer-electronics reviews, for example, while iVillage (a site focused on women) provided a section offering financial advice. Companies like these have achieved visitor conversion rates of 5 to 7 percent, double those of their competitors.

3. Avoid bleeding-edge technology: Sites that rush to acquire the latest software or to offer the coolest features typically have to correct flaws in the original design before long. By contrast, the top-quartile companies, rather than spending millions on, say, the latest piece of psychographic-profiling software, have concentrated on ensuring the flawless operation of basic product presentation features, customer service tools, and logistics. With the aid of disciplined outsourcing, these businesses have attracted and retained their customers while spending from 50 to 75 percent less than the average company in our sample.

The rules of the game

Accompanying such general imperatives are a number of specific rules for attracting, converting, and retaining buyers.

Attraction

Getting customers to a site for the first time remains expensive. You can save a lot of money by improving your targeting early on and making sure that any partnerships you enter into deliver large numbers of new customers at a relatively low cost. (Think of those terrible portal deals!)

Some untargeted mass marketing may be appropriate at the launch stage to generate name awareness. Even then, companies should avoid extravagant campaigns (such as the $2 million to $3 million that some dot-coms paid for 30-second spots during the January 2000 Super Bowl) and eccentric flourishes (such as efforts to put company-logo stickers on fruit or to persuade towns to rename themselves after companies). Similarly, a deal that lets your company place a logo and a hotlink on a general portal, at a cost of tens of millions of dollars over several years, will be less productive than a pay-for-performance deal in which an e-tailer might pay a $5 bounty to an on-line service like America Online for every referral culminating in a completed transaction.

Most critically, in the early stages of a marketing effort, companies must tirelessly experiment with and test their approaches. One of the companies in our sample tried out more than 500 different styles of on-line messages (for example, changing the text, the style, the color, and the format) to find the optimal marketing approaches for a number of different situations.

As the business refines its value proposition over time, marketing should focus on the pursuit of only those customer segments that have demonstrated their attraction to the business model. Affiliate deals with narrowly targeted sites (often exceeding 1,000 in number) and e-mail campaigns directed at segments grouped by purchase histories and demographic traits are 10 to 15 times more likely than banner ads on generic portals to attract prospects who click through to purchase screens.

Partnerships are the lifeblood of e-commerce companies. Best-practice ones invest a good part of their resources to create teams capable of negotiating contracts that pay partners only for actual results—unlike the large number of companies that signed partnership deals with no up-front, mutually acceptable performance targets. Such teams evaluate contemplated alliances on the basis of their strategic importance, their potential for creating value, and the likely difficulty of negotiating them. The teams also review completed deals systematically to determine which have attracted visitors most cost-effectively. Although this step may seem to be an obvious one, many companies, under the pressures of "Internet speed," are now echoing the rueful observation of one executive who said, "We’re doing deals a mile a minute and it shows."

Conversion

The strongest sites have conversion rates as high as 12 percent, as against 2.5 percent for average sites and 0.4 percent for poorly performing ones. Our data suggest that a handful of best practices underpin best performance.

First, the simpler the purchase process, the more visitors will be converted from viewers into buyers. Too many sites ask users for all sorts of data on their preferences and behavior. Free-PC, for instance, discovered that even the draw of a free computer wasn’t enough to persuade consumers to fill in lots of forms about themselves, let alone to look at a screen permanently half full of advertising.

Sites should initially request little more than payment and shipping details from customers. Some e-businesses have improved their conversion rates by over 50 percent as a result of simplifying the purchase process. Those that wish to acquire customer preferences and behavioral data must resist the urge to ask 20 questions off the bat. With adroit management, e-businesses can build rich profiles over time by posing one or two targeted questions during significant interactions, by collecting information during chat sessions, and by running sweepstakes that make the notion of value exchange explicit. Some sites give visitors access to more features (for instance, a daily newsletter or discussion groups) in exchange for a few extra pieces of information.

To make it easy to get to the point of purchase, good sites provide real-time assistance. Sophisticated product search engines, "call me now" buttons, live on-line customer support, and virtual customer-service agents (artificial intelligence software that can answer questions) are especially effective. One company that introduced live text-chat support increased revenue per customer by 35 percent. The ability to save shopping carts also strongly promotes conversion; simply by e-mailing customers who had abandoned them, some e-businesses have succeeded in making sales to as many as 30 percent of all such visitors. Many of these people thought they had bought something.

Second, best-practice companies make payment for transactions as easy, safe, and reliable as possible. They accept several payment methods (such as credit cards, debit cards, and checks) and obtain certification from consumer watchdogs such as TRUSTe. (Fewer sites than one might think bother to do so, even though consumer credit card fraud remains a major problem.) Successful e-businesses also actively promote consumer protection charters by, for example, prominently displaying information about no-questions-asked returns policies and about response-time guarantees.

Third, good sites seek to personalize their offerings, even for first-time visitors. At some sites, browser settings determine what visitors encounter; what they see when they go to Google, a search engine, depends on the country where they log in. Several US e-businesses use customer-profile-exchange technology, which gives Web sites information about first-time visitors by accessing on- and off-line databases.³

Retention

Perhaps our most important finding is the power of retention. A 10 percent increase in the rate at which visitors are converted into repeat customers drives a 10 percent improvement in the net present value of a company’s expected cash flows (Exhibit 4).

Early on, there was much talk about customer loyalty but insufficient investment in promoting it: e-businesses typically spent three to five times less to retain their customers than did equivalent brick-and-mortar businesses. Most of the money e-businesses did spend financed simplistic loyalty programs that awarded, say, $10 vouchers to customers when they had spent $250. Such programs significantly affect the surfing and purchasing behavior of fewer than 20 percent of on-line consumers. During the second half of 1999 and the first half of 2000, spending on loyalty programs rose. As of the first half of 1999, 45 percent of the companies in our sample had implemented relatively sophisticated loyalty programs,⁴ and 35 percent had invested more than $2 million each in retention-related tools; by the end of 1999, the figures were 75 and 65 percent, respectively.

The common trait of the companies that hung on to their customers was their reliability in basic operational execution. Their sites downloaded quickly; they responded to customer queries quickly; they delivered more than 95 percent of their orders on time; and they made it easy for customers to return or exchange purchases. One company that raised its on-time delivery rates to 90 percent, from 60 percent, cut customer churn in half.

Moreover, best-practice companies maintain current profiles of customers and send promotions keyed not only to their needs but also to their value. At the most basic level, this is about retaining and leveraging past interaction histories to ensure that the customer never has to answer the same question twice. More sophisticated companies seek to re-create the notion of the "personal shopper" by, for example, remembering key anniversaries of their customers and calling the best ones to make sure that orders arrive safely. If targeted marketing brings in the right people, and personalization makes it possible to make them the best offers, customer profiling ensures that a company focuses its resources on keeping its best customers happy. Best-practice companies also use e-mail to rouse dormant customers with attractive offers tailored to their buying histories—a low-margin, proven technique.

Finally, personalization is emerging as a driver of retention as well as conversion. A number of companies have deployed Web sites that dynamically personalize themselves in response to on-line customer behavior. Some sites, for instance, use the profiles of customers to determine which path through the site they take. Other sites use collaborative filtering to promote products that reflect user preferences suggested by past behavior.

But the path of effective personalization is open to relatively few companies, because of its complexity and the demands of mining large amounts of data. When you are trying to increase revenues and personnel at a 1,000 percent compound annual growth rate during the first two years of your business, a complex back-office system that costs perhaps $5 million and takes six months to implement may be something you can’t afford.

Sophisticated personalization is a nice-to-have, not a must-have, feature. The must-haves that help make e-businesses profitable permit you to find your natural customers efficiently, to offer them what they want in an appropriately segmented way, and to deliver it reliably. No surprise.

The e-performance scorecard

Data collection for the e-performance scorecard is rigorous: to ensure consistency, every classification into which data points are dropped has been defined sharply; all data are cross-checked with other sources when possible; and outlying data points are excluded. The e-performance scorecard comprises 21 indicators, which measure performance statically (at one point in time) and dynamically (over a period of time). These indicators are grouped into three categories—attraction, conversion, and retention—and then folded into the overall e-performance scorecard, a weighted average of the 21 indicators (exhibit).

The e-performance scorecard expresses two key dimensions of e-business: the efficiency of costs (for example, the cost of acquiring visitors to a site and of maintaining active customers) and the effectiveness of a site’s operations (such as conversion rates, the rate at which the number of customers increases, and customer gross margins).

Both individual companies and the e-business sector as a whole are rated by the e-performance scorecard. Both are plotted along a continuum, from best to worst.

Return to reference

Performing in Asia

The findings of a preliminary survey of Asian business-to-consumer (B2C) companies indicate that many of them may have avoided the excessive marketing expenditures characteristic of their counterparts in Europe and the United States. The cost of attracting visitors in the countries we surveyed—China, Hong Kong, Indonesia, Malaysia, the Philippines, Singapore, South Korea, Taiwan, and Thailand—is typically a good deal lower than it is in the United States and Europe (exhibit). Asian B2C companies also compared favorably with their US and European counterparts on several other operational metrics, including both customer retention and transactions per customer.

Yet Asian companies may have been too miserly with marketing expenditures and have thus failed to attract the kind of targeted visitor base that can easily be converted into a customer or membership base. The conversion rates of the companies in the McKinsey survey averaged 0.8 percent, far below the 2.3 and 2.5 percent of the United States and Europe, respectively. Of course, weak conversion rates mean that Asian e-tailers have relatively little money to spend on on-line advertising, and this shortfall may in turn be slowing the overall development of the B2C industry in Asia.

Operations issues aside, B2C businesses in the countries we surveyed face a much broader challenge than do their counterparts in the United States or Europe. The limited size of Asian national markets means that even best-in-class national players will face major challenges in creating significant value in the near term. To overcome them, companies must expand aggressively outside their natural markets while devising sound revenue models and continually strengthening their operational performance. To date, no company has managed to achieve this arduous but necessary growth trajectory.

—François Legendre, Mike Sherman, and Charlie Taylor

About the Authors

François Legendre is a consultant and Charlie Taylor is a principal in McKinsey’s Jakarta office, and Mike Sherman is a consultant in the Hong Kong office.

Return to reference