The search engines are pretty good at their jobs. This is especially true of larger, more established monster listings such as Altavista and Google. They have to be good, as they are in a constant state of war with search engine spammers (webmasters who attempt to artificially increase their rankings in search engines by unethical means).
You see, higher a site ranks in major search engines, more hits it receives. In many cases, hits directly translate into dollars. Thus, a web site which can, say, double it's hits can often double amount of money it makes.
What does this have to do with anything? Well, it's common knowledge that sites which do not show up on first three pages of listings in a major search engine may as well not be listed at all.
In addition, it's important that a site get listed on popular keywords. For example, far more people search for word "plumber" than "person who fixes pipes". while you might get a few visitors with later term, you will not get anywhere near as many as first.
Each of major search engines has different rules that it uses to rank web sites. Some engines want metatags, some prefer straight text and others want a mixture of both. Some search engines may be fine with dozens of keywords in a metatag, and others want only one. The list goes on and on - each search engine looks at different things in a page.
Why do they go through all of this trouble? They are attempting to determine what your page is all about. The theory is more a particular keyword (or phrase) is mentioned (and in more ways), more likely your page is about a particular subject. Thus, if "plumber" appears in text a few times (especially in H1 and H2 tags), in a metatag, an ALT tag and title, then it's likely your page is indeed about plumbers.
On top of that, search engines must protect against spammers. These are people who use various tricks to fool engine into thinking they should be well ranked. For example, a common technique a few years ago was to include very small, invisible text containing keywords. The visitors would not see this text but search engine would and thus would be fooled (the search engines figured this one out a long time ago and it no longer works).
When search engines discover a web site is spamming, their response is to either (a) drop site way down in rankings or (b) ban it entirely. If your site has ever been banned from one of big engines, then you completely understand how devastating it can be to be dropped all of a sudden.
But then again, getting to first can be so rewarding. It can mean difference between a thousand dollars in sales and a million. Literally. But how do you get to be first with a particular keyword in as many search engines as possible? One way is to look at other sites to see what they have done and, ahem, steal ideas (or just copy their keywords to your own pages).
But there is a wildcard in all of this, and that's simple fact that search engines use different rules to determine ranking of a page. One engine allows three keywords and will rank higher if it finds three, another might want those keywords to be near top of page, and still another might want them in a comment. The second engine (the one that wants keywords near top) might actually drop your rankings if it finds three keywords.
One of more common ways to handle problem of different search engines is to have different entry pages. Using this method, you might have a page which is perfect for Google, another which is exactly right for Altavista and a third which is made for Northern Lights. The problem with this, of course, is your visitors will be directed by each engine to pages which are probably not exactly right for human beings. After all, engines work even better with all of those fancy tables and lists which make your pages look so good. And, of course, this does nothing to prevent someone from stealing, uh borrowing, your keywords.