You've seen them. They're all over the Web. Some more prominent than others, but they're likely there, somewhere, revealing to the world how many "visitors" a site has had. I'm talking about the ubiquitous Web page hit counters...

...and why you SHOULDN'T use them.

I call them "ego strokers" for they serve no practical purpose other than to make a Webmaster feel good that his work's being looked at...sometimes. They usually fall far below expectations.

Aren't counters good measurement tools? Yes...and no. Yes IF you set them up correctly AND haven't fudged them.

Most counters are set up to measure "hits," not "unique visitors." Different things. A hit is a request for a server to deliver anything, like a Web page The same guy asking for the same page twice hits the server TWICE. Moving around a site can up that counter, and you'll probably NEVER see a 1:1 ratio between actual unique visitors and total hit count.

Is your counter doing hits or unique visitors? Those numbers probably don't look quite so good anymore... especially when you start getting technical. Each page loaded really counts as multiple server hits; page elements -- graphics, banners, buttons, etc. -- are really separate files and are all different hits. Technically. Some counters count them all.

Fortunately, the MS FrontPage 98 built-in counter counts a page request as just one. But if I click on the browser's reload button, that counter's going to count that too. Click. One. Click. One more. If I want to make it look like my site's getting lots of visitors, I could keep clicking away all afternoon.

But I don't even have to do that with FP98, or with any other counter for that matter. I can FUDGE it. If I wanted that counter to show 5 million hits, I could easily set it to start counting at, say, 5215097.

Which is the first point. You can't trust counters. And users are tuning in to that as they gain more experience.

Leading to the next point. Why does a user need to know those numbers anyway? Knowing the counts, a user may very well judge them too low (unimportant, useless site) or too high (the user's just one of the "numbers"). Smarter to just keep them guessing: act BIG and heavily used, but treat each visitor PERSONALLY and individually, like he's the most important customer in the world. Sort of like the "real world," isn't it?

Just leave those counters out.

But how do you count visitors then? That data's vital. Simple: Use a traffic analysis program, in the background, for your eyes only. All servers keep timed logs on just about EVERYTHING: how many hits each page (heck, each file) gets, how many unique visitors have come by, what domains and countries they're from, their entry pages, exit pages, etc. etc. etc.

Any good Web host provides you with a basic tool for reporting this data. Like Virtualis does (the Web hosting company I recommend to do-it-yourself businesses, described at http://www.e-factory.com/info/host.html). It has nice traffic reporting tools to help you numerically and graphically see what's up with your site. Other good vendors offer similar capabilities.

I prefer to dig deeper, giving me a clearer picture of what's actually happening: what people read; what's turning them off; where I win and lose them. Some reports tell me what parts of the site need work, and what to leave alone. And more. The data's right there, in the logs, ready for you to pick through; you just need a powerful tool to do it with. None of the stuff from the Web host's, not even Virtualis, dig this extensively. Not everyone needs them.

But if YOU find this detail valuable, check out WebTrends. It comes in various flavors, all built around the core WebTrends Log Analyzer.

It's NOT cheap at $280, but worth it. It's also part of more extensive WebTrends "suites," the Professional ($499) and Enterprise ($1,499). Yikes! Look at the suites only if you need what they offer, which is unlikely for the typical scope of a do-it-yourself site.

But you may not even need them if the tools from your Web host are sufficient for your needs. For most small businesses, they're usually more than enough.

As for Web counters, unless you're McDonald's, I don't recommend you touch them.