Net search engines rev up for battle

Ask Jeeves, the internet search engine, has come up with the best answer of all. Constantly asked by sceptics whether it would ever make money, the PG Wodehouse inspired business based in Emeryville, California, produced the clearest result this week.

Steve Berkowitz, the chief executive, announced that 2003 income was $22m compared with a $5.4m loss in 2002. Sales at the company were $107.3m compared with $65m the year before and in the fourth quarter alone Ask Jeeves sales were up 58 per cent to $31.8m. "Quarter four was another great quarter capping off a great year," said Mr Berkowitz.

But if you thought the Ask Jeeves results were impressive you should adjust your search criteria and ask the question about profits of Yahoo!, the rival quoted search engine that announced results two weeks ago. Terry Semel, the Yahoo! chairman and chief executive, said fourth quarter sales were $663.9m compared with $285m the year before and operating income for the full year was $295.7m compared with $88.2m in 2002. But in the fiercely competitive search engine world there are others even bigger and more impressive than this. Or at least we think there are. Google, by far the biggest search engine by its share of searches conducted on the internet, has yet to reveal any of its financial information.

Google's on-off flotation plans have kept the world's business press guessing for several months and the best estimates put its revenues at up to $1bn and profits of $300m.

Until its student founders, Sergey Brin and Larry Page, decide to go public - with the agreement of their financial backers Kleiner Perkins Caufield & Byers and Sequoia Capital - we will never know how profitable the business is.

What is clear, however, is that the search engine business model is proving increasingly successful as the market experiences explosive growth.

Using search engines is now the second most popular activity on the internet, after e-mail, and it is estimated that 550 million searches are performed daily on the web. With so many people using search engines they have become an advertisers' paradise. And where there are advertisers there are revenues to be had.

Whit Andrews, a research director of Gartner, the technology analysts, said: "If you do an online search then you are essentially taking a cable and projecting it into your forehead and telling advertisers more honestly than any other way what it is you want to know exactly."

The priorities for search engines now are to differentiate themselves in terms of technology, and the sophistication they can offer users. And they need to provide advertisers with ever more subtle ways of getting their message in front of millions of pairs of eyes.

The scramble to attract advertisers in this booming market is behind the ever more complex relationships developing between the search engines themselves. Some concentrate on simply marketing their brands and farm out the technology to another search provider. Some that were basically marketing machines are now going the other way and moving into the technology field. So, AOL, a marketing-led organisation, uses Google technology to provide the results for its search engine, AOL Search. Yahoo! was using Google as well until it bought its own technology company, Inktomi, last year.

Yahoo!'s move into technology shows just how fierce the competition is getting and how Google's dominant position is coming under increasing threat. Not to be outdone, Ask Jeeves also has its own search engine technology called Teoma. If you own your own technology, it is argued, the easier it will be to differentiate yourself in what would otherwise be a commodity market.

The charts show how the UK and US search engine markets are split by brand but also how dominant Google has been in providing technology for other search engines. What makes the search engine business model so elegant is that advertising can be highly targeted with pricing accurately linked to the number of people who actually see an advert.

Google, for instance, offers advertisers a system called Google AdWords. Advertisers get text-based advertisements that appear alongside search results which include a link to the advertiser's own site. Google's search technology matches a user's search criteria with an advertiser's text to increase relevance. The really clever bit is that advertisers pay when a user clicks on their advert. This makes sure the user is a quality "hit' in the eyes of the advertiser but also allows the advertiser to control its costs. It can set a daily budget for how much it wants to spend. Once it reaches its limit the advert no longer appears. It is the sort of one-to-one advertising that mass market mediums such as television can only dream of.

There are two other revenue generators. One is sponsored links. These come up as part of the search result and are clearly labelled as sponsored links. The other is paid-for placings. These deeply embed an advertiser's link in the search engine to make sure it appears when a very specific set of search criteria are used. "The search engines basically say 'we're going to show you an advert based on the information you've given to us in your search criteria and then charge the advertiser just to let it appear. We will then charge them extra if you click on the link," Mr Andrews said.

SPIDERS THAT WILL GUIDE YOU THROUGH THE WORLD WIDE WEB

Search engine technology starts with a piece of software called a spider or robot. It reads links across the whole of a search engine's index (rather than scouring the whole of the world wide web, search engines use their own indices containing 2 to 3 billion web pages). The spider categorises pages according to the key words of a search and builds a mini database in which it ranks its findings.

How efficiently pages are ranked becomes increasingly important as search engines try to differentiate themselves. There are three broad components that dictate ranking. First a text analysis finds key words then looks at a page image and compares that to the search criteria. Second is a popularity element that measures how many links that page has from other pages on the web. Finally the spider looks at how many times a page has been clicked on before from previous searches. Other refinements look for how many links a site has from other sites in the same subject area to help establish popularity.