E for Two

E for Two

A tale of two newcomers to the e-commerce arena. One built its own computing platform; the other bought the whole kit and caboodle from a single vendor.

Which strategy is best? The market will decide.

In this story, you'll learn

* Ways to decide if e-commerce makes sense for your company * The relative strengths and weaknesses of proprietary versus single-vendor platform strategies * How to leverage the technology to satisfy your e-customers Decisions, decisions. Life is an endless string of choices to be made. And few decisions are as momentous as the ones now facing companies entering the hurly-burly of electronic commerce.

For e-comm start-ups, everything hinges on the selection of a computing platform that will see their company through its salad days to the million-hits-per-minute stage that dreams are made of. The challenge is to build an enterprise platform that will scale up easily but not break a start-up's modest finances. The three basic choices are to build your own (which can be extremely expensive to create and maintain), choose a single-vendor platform (which ties you to the fortunes of one vendor) or use a mix of off-the-shelf products (which makes you a systems integrator). There are no easy answers.

And with the crazy pace of Internet time, the market will rule quickly on the wisdom of your approach. But even if you've made the wrong decision, there's no turning back: Once you've invested in a particular strategy, you'll have to stick with what you've got. Otherwise you're out of cash, out of time and out of luck.

Take, for example, Buyonet International AB and Holt Educational Outlet (HEO), two relatively tiny companies that took very different approaches to targeting the online consumer market. Buyonet, a Swedish electronic software retailer, and HEO, an educational toy and teacher supplies retailer, were both attracted by the explosive growth of the U.S. Internet commerce market, the business-to-consumer sector of which Forrester Research Inc. in Cambridge, Mass., forecasts will reach $US108 billion in 2003, when 40 million households are expected to shop online. Both companies saw estimated revenues of between $US1 million and $US3 million in 1998. Buyonet currently has 20 full-time employees (and uses contractors heavily); HEO has 40.

But despite the similarities, the companies' top executives have strikingly different philosophies. Buyonet CEO Freddy Tengberg chose to roll his own e-commerce solution, hiring a team of contract programmers to develop a sophisticated Unix system. Tengberg believes his system's robustness and flexibility will give him a sustainable competitive advantage. Holt Educational Outlet CEO David Lord, on the other hand, opted for an all-Microsoft Corp.

BackOffice solution, and he is convinced this platform will be easy to maintain and less expensive over the long haul. Time-and the marketplace-will judge which strategy was better. In the meantime, CIOs-whether seasoned players or newcomers in the online commerce arena-will benefit from following Tengberg's and Lord's divergent journeys down the e-commerce path.

To E-comm or Not to E-comm?

The first question facing any company considering this path is whether to get into online commerce at all.

As far as Tengberg was concerned, there wasn't much doubt. Buyonet was founded in Gothenburg, Sweden, in 1997 for the express purpose of selling and delivering software online. And despite his location, Tengberg wanted a piece of the burgeoning U.S. online software market, forecast by Forrester Research to reach $US3.2 billion by 2003-by far the biggest segment of the online consumer market. "We need volume," says Tengberg. "Volume exists only if you sell to a large group of people." Tengberg soon saw that being located in Sweden did not have to be a disadvantage. The online world is indifferent to nationality - so long as the commerce site has value for the user. Tengberg originally thought Buyonet would first target Swedish users and then expand into the rest of Scandinavia. But, he says, "It took only five minutes to see that this was an old-fashioned way of thinking. There are no borders on the Internet. You're nowhere and everywhere at the same time." So it was settled. Buyonet would be a global player. But the company was entering a market already populated by leaders like (previously and In order to have a realistic chance to compete with those market pioneers, Tengberg would need an extraordinarily robust commerce infrastructure that could support multiple currencies and languages.

For Lord at HEO, the situation was not as clear-cut. HEO was a division of The Holt Co., which was not a start-up-for over 20 years it had supplied chalk, erasers and other essentials to schools. (Holt's other two divisions sold meeting room furnishings and projection and videoconferencing equipment to schools and businesses, respectively.) When Lord joined Holt in 1997 as CFO, he had one overarching goal: to champion growth. To do this, he had his sights set on HEO. In the past decade Holt had expanded into educational and specialised toys, opening a discount retail store in Waltham, Mass. And as late as February 1997, Lord thought growth meant opening new HEO stores. He came on board intending to open six new retail stores by the end of the year.

But he quickly realised that HEO customers - time-pressed baby boomer parents looking for educational toys for their children - were a slam-dunk fit with the highly educated, highly paid demographics of the average Internet user. "Parents who have too little time but don't want to settle for the quick plastic gizmo at Toys "R" Us - that's our target market. So we felt the fastest way to get to those customers was on the Internet," says Lord.

With e-commerce as the goal, Lord soon added CIO duties to his job (he has since been elevated to CEO of Holt) and turned his attentions to the pitiful state of the company's technical infrastructure. While many established companies adapt their existing systems to e-commerce, Lord opted to rip out the tiny Novell network running a legacy financial system. That decision, at least, was an easy one. "The computing platform was terrible. It wasn't going to support growth," says Lord. The old financial system was so unreliable, in fact, that Lord did not even convert the old data over to the new one, choosing instead the arduous task of repopulating the new system with financial data by hand.

Build or Buy?

Buyonet was a start-up; HEO part of an established (if tiny) company. But both were beginning their e-commerce endeavours with a blank sheet of paper - free to pursue any technical architecture within their humble means. (Buyonet was backed by several million dollars in venture capital; HEO was funded to the tune of a few million by Holt.) The choices facing both companies: build, buy or integrate products from multiple vendors. Buyonet and HEO elected to build and buy, respectively.

After a somewhat peremptory search of e-commerce software products, Tengberg decided to build the Buyonet system from scratch. In some ways this is the most ambitious - and certainly the most resource-consuming-choice he could have made. But Tengberg believed his company's distinctive requirements would not lend themselves to any off-the-shelf product.

For one thing, unlike many of its electronic software distribution (ESD) competitors, Buyonet's business model was to allow customers to download the software first and then pay for it. (Although shareware companies have been operating this way for years, most software sites on the Internet require payment prior to download.) Once the download had occurred, the Buyonet system would automatically check that all pieces of the program had been transferred successfully to the consumer's PC. Only then would payment be required. The software program would be disabled until the user entered a special password sent to him via e-mail.

Market research in countries other than the United States indicated individuals from some cultures would not accept the idea of paying for something they had not yet received. Allowing the download to occur prior to payment also let customers avoid the hassle of wondering what to do if something went wrong with the download - as often happens. "Do you press cancel and hope for the best, or do you continue? If your Internet connection is suddenly cut off, what do you do?" says Tengberg. "We wanted to make sure it was securely on the user's hard drive before we asked for payment." Because of the company's global emphasis, the Buyonet system also had to support multiple languages - including English, French, German, Portuguese, Russian, Spanish and Swedish-and multiple currencies (more than 20). The Buyonet team wanted the system to be able to detect what language the operating system of the incoming browser was in, in order to interact seamlessly with the customer in his own language. "No one had the technology to fulfil our needs," says Harry Klintebring, the U.S. marketing manager for Buyonet.

So the company contracted with a software development company in Riga, Latvia, where there is a surplus of qualified software engineers. In four months, those programmers built a sophisticated and flexible platform for the company to sell more than 120,000 software titles (all of which reside on the company's or on partners' servers). The team used C++ and Perl to create a complex system that runs on Sun Microsystems Inc.'s Solaris Unix machines. The main system has eight Unix-based subsystems - for delivery, antifraud, reporting, sales and database analysis, payment, special activities (such as promotions and affiliation programs), Web site and systems maintenance-that work off a TcX DataKonsult AB MySQL database (see "Buyonet's E-Commerce Infrastructure," Page 49).

One Sun Enterprise 250 server and one Sun Ultra Sparc II server reside in Gothenburg, and another Ultra Sparc II resides in New York City. Having servers in two locations keeps problems with the site's Web traffic - which runs at about 10,000 unique visitors per day-to a minimum. Scaling up to serve more customers is a simple matter of adding new server hardware.

Users have five options for payment: via credit cards protected by secure sockets layer (SSL), via credit card using the secure electronic transfer (SET) protocol, by faxing credit card information or with a bank money order or check. (The last option is available only in the United States.) Buyonet is connected by a separate ISDN link to the National Westminster Bank PLC in London for instant payment processing. The customer receives a reply within 30 seconds - and usually more like 10 seconds - as to whether her form of payment has been accepted.

To date, Buyonet has spent roughly $US2 million on technical development of its site ( and only about $US40,000 on hardware. As the HEO example will show, this is indeed a more expensive approach than buying a single-vendor, off-the-shelf solution. For that reason alone, more and more companies are choosing to go the single-vendor route, says James McQuivey, analyst for online retail strategies at Forrester Research. Although proprietary technology gave e-commerce pioneers an early lead, that's no longer necessary. "I can have all the features without knowing a single line of code.

The drawback to developing your own technology is keeping all those programmers on staff. Once you have that snowball rolling, it just keeps going," he says.

But Tengberg fiercely defends his choice to custom-develop his company's commerce technology. "[The custom system] allows us to build in unique services such as a buyer bonus program. [Packaged solutions] don't have those features or our support for multiple languages. Maybe the vendors could develop it on their schedule, but not on ours. The winners are going to be those who act quickly and are most adaptable," he says. And because it used contract programmers for the bulk of the development work, Buyonet will not have the burden of having numerous programmers on salary.

Indeed, although Tengberg says it is too soon to quantify an ROI for the system - he expects to be able to do so next year - he does go so far as to say that having built a proprietary system gives Buyonet lower transaction costs (such as for order fulfilment) than other options. For example, companies that used licensed e-comm products or outsourced services typically have to pay a portion of the revenue from each sale to the vendor. "With margins shrinking, we will be able to make money out of each transaction while others are in a break-even or loss situation," says Tengberg. Not surprising, Buyonet has applied for a patent to protect the proprietary technology that it worked so hard to create.

The Single-Vendor Approach

Buying into a single vendor is also admittedly risky. Says Forrester Research's McQuivey, "Most CIOs are not going to be comfortable trusting their fate to a single vendor." But that is just what HEO decided to do.

After trashing the old system, Lord went to work revamping the technical architecture from the four walls out-starting with 100-megabit wiring and state-of-the-art PCs and then installing Microsoft's BackOffice. HEO runs the enterprise edition of Microsoft Site Server distributed over six ALR 6x6 servers, which are combined into a single virtual server via a Convoy cluster.

The database management system is Microsoft's SQL Server 7.0.

The heart of the e-comm platform is a Great Plains Software Inc. Dynamics C/S+ financial system, a state-of-the-art, object-oriented system that is integrated with the Web site. Orders come off the Web site and flow automatically into the financial systems for processing and fulfilment (see "HEO's E-Commerce Infrastructure," below); most online vendors still use human intervention to bridge the gap between their Internet front ends and their back-end transaction processing systems. Customers can pay with any major credit card or elect to use the toll-free number. Payment by check is not an option.

Buying the all-Microsoft solution was a way of cutting to the chase. "We felt like, Let's leverage this technology. We're a small company - we don't need to spend our time integrating different products," says Lord. "Microsoft may not be best-of-breed for everything, but it's the best for us." Lord acknowledges that it may cause some discomfort to be dependent on one company: "You are at their mercy. You have to stay in tune with what they're doing; where they go, you must go. Changing in midstream would be very difficult because you're tied to their solution." Nevertheless, Lord is not bothered by that or by the fact that a single-vendor solution may be less scalable in the long run, and he notes that he has an excellent relationship with Microsoft.

(Indeed, he was the keynote speaker at Microsoft's E-Commerce Day in December.) And it seems pretty clear that Microsoft will dominate well into the next century.

As for the prospect of building his site on proprietary technology, Lord scoffs. "It's all a matter of cost and profitability." The early e-comm players had no choice but to roll their own. But as Internet commerce becomes business as usual, the companies that are not stuck maintaining a complex home-grown system will be better off than those that are, he says.

Lord is certainly right that buying a single-vendor suite is cheaper in the short term. Compared with Buyonet's $US2 million plus, HEO spent only $US500,000 to build its site. This breaks down to about $US150,000 for hardware, $US50,000 for software, $US100,000 for miscellaneous costs and about $US200,000 for the Great Plains financial system.

And the transaction costs of having a totally integrated Web site are enviable. Prior to launching the integrated Web site, HEO limped along on its old platform, getting orders online and manually inputting them into the back-end systems. The per-transaction cost under that architecture was $US2.25.

Under the integrated system the cost has dropped to 50 cents per transaction.

All told, the system paid for itself by the end of the December 1998 holiday season, Lord says.

Low technology and transaction costs have freed Lord to concentrate on giving users more ways to find and buy toys. For example, there are lists of hot toys, the top 25 best-sellers and best deals in addition to a whiz-bang search engine. "Customers want to have as many choices for finding the best toys as they can," says Lord.

Every page on the HEO Web site ( is dynamically created by the system at run time according to a cookie-based user profile. In addition to 20,000 educational toys, the site contains a number of instant polls such as "Do you read to your child every day?" and "Have you taken your child to a museum in the past year?" More than 35 per cent of the site's roughly 25,000 daily visitors take a few seconds to participate in a poll. The more polls the user answers, the more the Holt system knows about the person. It then serves up personalised pages containing toy suggestions based on what it has determined are the person's interests, a technique famously debuted by

One downside to Holt's Web page is that it cannot be accessed by people who have elected to turn off cookies. Lord dismisses this group as representing only .005 per cent of current visitors, a number he expects will drop even lower as consumers' Internet privacy fears dissipate further.

Holt is winning a lot of attention these days, riding the explosion in online toy sales. Forrester Research had to revise its initial 1998 estimates for the online toy market from $US20 million to $US70 million. This is still only a drop in the bucket of overall toy sales, but online toys is one of the biggest growth areas on the Internet, according to McQuivey. "By 2003, we estimate merchants will end up selling $US1.5 billion in toys online-that's a compound growth rate of 85 per cent per year," he says.

Last December the excitement was so high that HEO almost capsized in a tidal wave of Holiday Hoopla (the name of the company's Christmas shopping program, which offered free overnight shipping and gift certificates with purchase). Then Money named the site the best for buying toys online.

Overnight, user traffic doubled to 50,000, and the number of transactions went from roughly 750 to 2,000 per day. Since Lord had thought big from the beginning, he was prepared when success came knocking. He simply added servers - at the rate of two per week throughout the holiday season - to handle the traffic. These are nice problems for an e-commerce company to have.

And these happy problems seem likely to continue well into the future. Says Forrester Research's McQuivey (with a nod to Ross Perot), "The Internet is the new source of the giant sucking sound, sucking people into online shopping.

With all these people online, more merchants are online. It's a growth spiral." An E-Comm Primer Still an e-commerce novice? A new book can help you polish your online strategy.

Patricia Seybold, founder of the Boston consulting firm Patricia Seybold Group, has written a lively and fast-paced book that should be required reading for everyone in corporate America. Published by the Times Books division of Random House Inc. (1998), How to Create a Profitable Business Strategy for the Internet and Beyond is an excellent primer on e-commerce. The book tackles the subject from the viewpoint of the all-important customer, pointing out holes in even successful companies strategies. Seybold lays out the five steps to e-commerce success and the eight critical success factors to reaching that point, with detailed case studies of 16 famous and not-so- famous companies.

Seybold uses personal anecdotes about her struggles to get excellent service to great effect. The case studies on companies ranging from American Airlines Inc. and National Semiconductor Corp. to the National Science Foundation and Tripod Inc. are clearly laid out to help readers quickly absorb the message. For example, technical details concerning the profiled company are in a different typeface and set off from the rest of the material so that those who prefer can easily bypass the techie stuff.

The book is packed with take-away messages and lessons learned from those who have done it. There's something for the established retail company worried about channel conflict as well as for the starry-eyed entrepreneur dreaming of e-comm fortunes. In a speeded up Internet world, this book won't waste your time.

- L. Paul

Same Expert Advice

Patricia Seybold weighs in on the e-comm strategies of Buyonet International AB and Holt Educational Outlet In her new book An e-comm primer, above), Patricia Seybold offers advice to companies that sell online. Here, she spares a few words of counsel for our profiled fledgling e-comm companies.


Seybold believes Buyonet is facing stiff competition from the likes of and, which made it into the marketplace first. There's no joy in trying to out-Amazon an, a position in which Buyonet may well find itself. The cure, Seybold says, is for Buyonet to differentiate itself from its competitors by offering support for the products it sells.

Buyonet could offer a central support location that would link seamlessly into the software vendors product support pages. Buyonet should also create a sense of community by letting users help each other with common software products in a bulletin-board or forum format.

Holt Educational Outlet:

Seybold says HEO should make as many deals as it can to link reciprocally to other toy and consumers goods vendors Web sites. The larger online toy sellers such as and should be more than willing to hand over a large commission on any products purchased by users coming from the HEO site.

The reason? It costs large companies a hefty amount on the order of $US40 to $US50 to acquire each new customer through advertising or direct mail. Under these conditions, HEO could ask for a commission just under that amount, Seybold says.

HEO already scores high on creating a sense of community among its users, which Seybold says shows a highly evolved Web company. Links to articles on buying toys for disabled children, for example, and a special relationship with famed paediatrician William Sears earn Seybold's thumbs up.

- L. Paul

Editor's Note: At press time, it was confirmed that Holt Educational Outlet (HEO) has been spun off into a separate company called, with David Lord presiding as CEO. As mutually agreed upon by Lord and Holt management, the new company focuses exclusively on selling toys; its retail outlet is now called, and the Web site's new URL is Holt now concentrates on its two other divisions, which sell furnishings and equipment to schools and businesses.

Join the CIO Australia group on LinkedIn. The group is open to CIOs, IT Directors, COOs, CTOs and senior IT managers.

Join the newsletter!


Sign up to gain exclusive access to email subscriptions, event invitations, competitions, giveaways, and much more.

Membership is free, and your security and privacy remain protected. View our privacy policy before signing up.

Error: Please check your email address.

More about ALRAmazon.comAmerican AirlinesBeyond.comBoston ConsultingeToysForrester ResearchGreat PlainsMegaBitMicrosoftMySQLNational SemiconductorNovellRandom HouseSeyboldSun Microsystems

Show Comments