Rather than spend a lot of money merely reproducing its core business on the Web, insurer CGU decided to think smart and use the Internet to maximise value. The result, nine months later, is Bluecycle, an auction site for salvaged goods that was dreamed up by a project team pulled from all corners of CGU to develop business ideas for the 21st century.
Like all big insurance companies, CGU owns a massive amount of stock by dint of the claims it pays out on. When a car, for example, is written off and a claim made, the insurance company becomes the owner of that car. Before Bluecycle was set up, the car would typically be sold off for a nominal sum to a jobber, who would pass it on to a retailer, who in turn would sell it to a customer.
"We asked the question: what do we do with this stock to maximise value?," says Sean Egan, CEO of Bluecycle. "At the time, the answer was 'very little'. But we realised the Internet provided us with an opportunity to cut out some of those jobbers by auctioning the stock online."
According to Egan, the business model had to be revised radically in the light of experience. "We started out from the basis that successful online auction houses exist and decided to ape them," he explains. This meant Bluecycle started out expecting to do business with private customers rather than organisations. Early trading experiences encouraged a rethink.
"Our focus is moving to business-to-business," says Egan, who gives two key reasons for the change: "First, the B2C market is very crowded; and second, for the type of salvage CGU specialises in, the natural outlet is business. Salvage lends itself to business because of the types of goods and quantities involved. If goods are water-damaged, say, a dealer is happy to buy them where a consumer is not."
Unlike other auction sites, Bluecycle doesn't just introduce buyers to sellers and leave them to get on with it. It brokers the deals, taking responsibility and risk for the financial transaction. Once a buyer has paid up, the seller does not receive payment until the buyer has the goods.
"We wanted Bluecycle to be a site users can trust," says Egan. "'Powered by CGU' has become a large part of the brand. People have the security of knowing they are dealing with one of the top 200 blue chips, not a fly-by-night. For CGU, this is part of a long-term play. We're a sustainable business, not out to make a few quid on the stock market." Indeed, CGU's clout was crucial in winning financial backing for Bluecycle. Banks would be unlikely to underwrite a similar arrangement for a start-up.
Normally, the intended functionality of a project tends to define how and when it is implemented. Egan, however, believes Bluecycle had an overriding reason for moving more quickly than that: "The usual thinking runs, 'We need this set of components and therefore can't deliver until April.' But you have to trade this objective against the experience of being in the market. Personally, I think it's better to be learning about how it is to be in the marketplace than spending time getting everything perfect. You can kid yourself about getting it right."
He points out that a lot of blue chips are quietly launching sites. Only after learning a few things on the way do they slowly start to advertise their presence.
This resolve to learn from the market means compromising on methods that are written in stone for traditional IT projects. "If I was being judged by the benchmarks of a classic systems implementation, I'd be taken out and shot for what I've done," jokes Egan.
He believes one of the key differences with e-business is that more time needs to be spent getting the software right than building the perfect technical architecture. "You can argue that scalability issues deserve as much attention, but I'm happier dealing with those as I go along than working with software that is only 40% right. It's important our customers have a reasonable experience from day one."
Egan's experience as a consultant with IBM helped him see development from two perspectives: the techie's desire for elegant solutions and the user's need for functionality. "In other development environments there tends to be the attitude that the customer will just have to live with the software, but in the e-space that is wrong. The customer relationship is paramount."
Egan envisaged having to achieve distinct goals - namely, creating a brand, opening up a market, organising operational processes and individuals to achieve goals, and building a technical infrastructure. He decided to deliver sets of functions by hiring experts in branding and technical solutions. MNPT worked on the creative side and Quidnunc worked at Bluecycle's offices for 10 weeks to launch the site.
"One of the first things I asked Quidnunc for was a project plan - but I didn't want to spend time on change management control, rewriting the plan each time it was rejigged." One of the advantages of working with a small team in a concentrated way is that everyone can physically be brought together. "We're set up separately without constraints in a space where we can do what is necessary," he says.
The only way Egan could deliver on his relentless timetable - his orders were to set up the Web site in 90 days - was to take a ruthless approach: "I've been pretty rigid about 'here's a specification and here's what I'm delivering'."
Egan acknowledges there is a desire to delay releases until everything is perfect, which would be perfectly appropriate for traditional big systems development with its longer-term focus, and to organise development phases into bigger tranches. But one compromise he won't accept is a failure to come up with releases. "We want improvements out there, even if it's only 50%," he says.
A crucial phase of the project was the 'soft launch' of the site to friends and families in February. This helped identify unacceptable shortcomings in the customer-facing software. In the grid that displays goods, for example, the inital version didn't bother to put a tick against what was in stock. "That was no good to our users," he says.
Egan recognises the site is not perfect, but accepts that incremental improvement is a fact of life. The Bluecycle technical team is currently working on launching new functions at the end of May. Egan admits there is pressure to delay until June but believes there is more benefit to the customer in delivering something in May.
On the upside, because adding WAP functionality proved relatively easy, it was released early. "WAP is perfect for auto salvage dealers," says Egan. "They don't want to spend the day in front of a computer waiting to rebid for that Jag convertible. With WAP, dealers can stay connected and rebid while out and about, wheeling and dealing."
How Quidnunc did it
Because of the nature of the business, the architecture for Bluecycle could be free-standing and the main framework was defined by the auction-house functionality. Out of the out-of-the-box products on the market, Quidnunc selected Moai, chiefly for its openness and Unix support.
Unix was the obvious pick for scalability and performance, says Martin Wilson, project architect. "For B2C, where potentially anyone in the world might log on, you need the extra performance, although we might use NT for B2B." Quidnunc picked Solaris and an Oracle database with an IBM WebLogic application server.
Moai comes with a sample, ready-to-go auction site including a database of products and the process of bidding and notifying bidders. But many features had to be customised or added for Bluecycle. For example, because Bluecycle offers various methods of payment, the technical team had to develop its own process to kick in once a customer has won a bid. "Phase one of the project included writing objects for payment, shipping details, and how to invoice," explains Wilson. Writing the presentation layer and customising all objects took 10 weeks.
The phase leading up to launch had three key stages: business requirements, technical design and user testing. User testing included trying out the content map and usability-testing the navigation. Quidnunc followed a three-month lifecycle for development and release, and used time-boxing and prototyping to stay on schedule.
Quidnunc uses an N-tier architecture to build Web sites because of the flexibility it gives for adding and tuning functions. Because a middle-tier application server is used to make all the calls to databases and create presentation pages in different formats, it's just a matter of writing Java components and slotting them in, rather than having to tweak HTML or tune graphics. This made adding WAP functionality, for example, very simple.
Wap's a snap
WAP is still a primitive standard. The bandwidth available between WAP phone users and the Bluecycle servers is just 9,600bps - the speed of modems around five years ago. But because the Bluecycle application needs only small amounts of data (informing bidders their offer has been surpassed and inviting them to rebid), performance is not affected, according to Nick Maxwell, partner with Quidnunc.
And making a site WAP-ready is a piece of cake, as Martin Wilson, project architect, explains: "In effect, anything you can do online you can do via WAP. It's simply a case of using Wireless Mark-up Language [WML] to display data on WAP phones and scripting objects on the application server in JHTML to make the calls. The back-end databases are thus insulated from having to know about and deal with a new type of device. "Once you've done that, the WAP device will work with application logic and legacy systems. In theory, if you've got a Web site that integrates the application to a back-office system like SAP, it's extremely easy to turn it into WAP."
The only real issue is that the screen is really tiny and you can't send rich content, adds Wilson. Users without WAP phone can still be informed of the need to rebid using the basic SMS messaging, but would then have to get online via a computer.