How to build Web business
The world is littered with the debris of failed dot-com companies. Some launched on the stock market and still managed to go bust, while others did not even get that far before frail revenue streams and ill-conceived business models dragged them down. But, among this carnage, some Internet companies are surviving. Some are even thriving, since the state of the market has forced out their less cunning competitors.
On the face of it, now might seem to be a foolish time to launch an Internet business, but that is not the case. If an economy can support new businesses in other sectors, it can support them on the Web. But whether companies are adding a Web presence to an existing business or creating a pure Internet subsidiary, they should tread carefully. A good business plan is essential, as is a thorough understanding of where the pitfalls lie.
It is important to research the potential audience before site development is begun. This means finding out what visitors want, how they want it and when they want it. And, of course, what they are prepared to pay for it. It is not always necessary to commission market research; careful scrutiny of other, successful sites in the same market sector is informative, as is a good look at the unsuccessful ones. Firms should ask themselves if the market is large enough to support a new venture, whether they can do a better job than competitors and whether it can be done while still making a profit. If not, it is probably time to rethink the plans.
If the boom and bust of the last couple of years has taught industry one lesson, it is that conventional advertising models are usually not sufficient to keep a site afloat. The standard, free-access Web site can use advertising to help cover its costs, certainly, but there has to be something more to maintain profitability. This fact was largely hidden by the vast amounts of venture capital being invested in dot-coms. Now that has gone, the truth is more apparent.
There are several options here. It is possible to opt for an ad-supported, registration-based site, but with free access. This would give some information about visitors, along with the all-important email address. This sort of information is likely to be appealing to advertisers, so the site would be able to attract more money than less targeted sites.
Alternatively, the site could charge for entry, although the number of people who are willing to pay for access to sites is low and will remain so unless there is a concerted effort from the big publishers to move to a paid-for model.
Some firms do deals with other companies to forward customers to their sites, charging them for the privilege. Cost-per-click deals are still undervalued at the moment and the organisation will have to be careful that it is not giving away branding when it is only being paid for traffic generation, but solid sales leads are hard to come by and companies are willing to pay well for them in the right circumstances.
Of course, many sites will simply be designed to provide e-commerce shopfronts for existing offline business, or to provide information to customers and staff. In these cases, the value of the sites is likely to come from increased sales or increased efficiency and competitiveness, rather than advertising revenue.
Companies should be careful that their site design does not unintentionally exclude members of their target audience. Approximately 85 percent of consumers use the Internet Explorer browser, though not always the latest version. It may be tempting to cater only for them, to make site design easier. However, this cuts out 15 percent of the potential audience simply because an organisation does not want to build browser detection and graceful redundancy into the site.
If the target market for a site is primarily home consumers, firms might try this test. Take home a notebook PC with a V.90 modem, set the laptop display to 800x600 and, after completely emptying the browser cache, access the site over a dial-up connection. How long does it take to load? If it is more than 20 to 30 seconds the site will lose business. Is it possible to browse with images turned off? Can text-based browsers access the site? If not, there is room for design improvement.
It is usually wise to avoid proprietary site components, including Flash, or to provide a standard alternative to users who do not have, and do not want, extra plug-ins. Firms may think a 500kB movie is fundamental to their corporate image, but many users are likely to get bored with the download and go elsewhere, probably straight to a competitor's site. Firms should not let design get in the way of functionality and navigability.
Companies should also beware any analyst who, before even looking at the business, says it needs, for example, three dedicated servers running SQL Server, an Access front end and a dedicated mail server. Unless the company already has a serious Web presence or is a very well-known brand, it is best to start small and upgrade as necessary. Consider Linux as an alternative hosting solution to Windows NT or Windows 2000, because the software is usually free and very reliable, and try MySQL and PHP instead of the commercial alternatives. They are efficient, powerful and there is plenty of free code available that can be tweaked to suit individual needs. Support is good too, with news groups and Web forums responding to problems very quickly.
Companies should also avoid the use of hosting services outside the UK, unless they specifically want to build international traffic. The extra hops add to the download time of a site, and hosting costs are now quite similar on both sides of the Atlantic. However, it is often not cost-effective for a firm to host its site itself because the task involves a large amount of work compared with using an FTP and Telnet connection to someone else's server.
When a Web site has been set up, it makes sense to monitor it to judge its performance and see if any improvements are required. However, Web server logs are difficult to interpret. Whether firms use Analog, WebTrends or some other analysis package, they should be aware of the limitations. A unique IP address is not necessarily the same thing as a unique user, for example, and no log file will say when a user actually left the site. Site impressions are not definitive either, even for advertising-supported sites. Search engine spiders can artificially inflate the statistics unless they are specifically excluded from the figures.
Fundamentally, what matters is the number of active users visiting the Web site. If 10,000 people visit, but total page impressions are only 15,000, there is something wrong with the site design or implementation. If it is an e-commerce site, but only three of those 10,000 visitors made a purchase, then excluding high-cost, high-margin operations the site is doing something wrong. Managers should understand these fundamental statistics and should not be distracted by the other, less relevant numbers that log file analysis packages can generate.
Approximately 80 percent of visits to Web sites originate from search engines, so this should form the focus of any marketing campaign. There are firms that will, for a fee, tweak a site using meta tags, entry pages and so on, to attain better rankings in specific search engines, but ultimately this is something that is probably better done in-house, as staff will have a clearer idea of the intended goal.
Firms should also make good use of email. Companies are realising that targeted email is effectively direct mail, and so can come out of the direct marketing rather than the advertising budget. Return on investment is usually pretty good, so firms should be marketing to their own databases of email addresses, to let customers know about offers and items on the site, and perhaps to sell advertising space on their email newsletters.
Downtime can be expensive if firms lose business and customer goodwill as a result. Companies should not rely on the hosting company to make backups of their sites, and should implement a full, hierarchical backup structure with fast restore. Remember, if the site is down for more than 12 hours, it could be removed from some search engines and hard work will be wasted.
In some cases, replicating live databases remotely is not practical, in which
case it is important to ensure the hosting company is doing it. Even traditional
firms will struggle to survive major data loss. For an Internet company, such
a loss would almost certainly be terminal.