The key to obtaining great results from a poll is a question everybody has an opinion about, one they want to share. I goofed in this regard with a poll running on my site. I asked visitors their opinion about using HTML in business email. Response has been so low, it's clear not many are concerned about it.

Still, even these limited results tell me something of my visitors. That few are interested in this question is part of what I learned. While this may seem a trivial point, it does apply. I will be hesitant to say more on this subject, for I want to produce content readers and visitors want.

The only surveys I've tried on the Web have been offered via "STAT News." It's difficult to get a significant response in this format. For one thing, we're all quite busy and it takes time to figure answers to questions. Additionally the survey taker must reply via email; many are hesitant to do so for fear of their address being misused.

One way to obtain anonymous response is to put a survey up on your site. http//SurveyMonkey.Com makes this easy to do. While I haven't tried it yet, I will. Again, the more interesting the topic, the more input you'll get. And the quality is likely to be correspondingly better.

An honest poll or survey seeks a random sample, a small part of the whole. It's tough to obtain. And few get it right. With a survey or poll in your newsletter or on your site, the sample you get will *not* be random. Why?

Those responding may be more aggressive types, not representative of the whole. Or you may have phrased your questions in such a way as to turn off readers and visitors of one sort or another. Composing questions which all or even most will understand as you meant them, is a real challenge.

If you get a relatively large response, results can be taken more seriously. This is not the usual case, however. The mode is to get a few responses, but not nearly enough to consider results representative of the whole.

I personally am willing to extrapolate and say that any one response represents that of nine others, had they taken the time to participate. But I won't go further. Thus given say 4000 subscribers and a response of 100 to a survey, you may be able to draw some reasonable conclusions about 1000 subscribers, but nothing much about the other 3000.

If you require an email address, you will turn off those unwilling to give one. Input will come only from those who will, and those who give a phony address. Either way, any randomness in the sample is gone. And results are likely to be meaningless.

Offering a freebie will almost always kick back on you. At least some will participate just to get the freebie. Some will even try to twist results, just for the fun of it. The greater the perceived value of the freebie, the muddier the results are likely to be.

Forums, bulletin boards, and chats are terrific. But they don't work well without sufficient traffic. On the Web, image is everything, at least initially. Nothing I can think of at this writing can destroy an image quicker than a "dead" forum in which the last message was posted a month back. (And don't think for a moment you can fake it; the only luck in trying this is all bad.)

Further, it takes time to manage the input. To even consider letting stuff flow without a moderator is foolish. One solution is to find a volunteer to moderate, someone interested in the topic who can benefit from the exposure. It's easy enough, then, to drop in now and then and add a comment of your own.

Email is the most effective tool I have found for building demographics. On my site and throughout my newsletter, I invite comments and questions. In fact I beg for them. In answering, I have an opportunity to generate further feedback. More important, it allows me to demonstrate expertise and make the first connection in what may grow to be a significant relationship. And from every message, I gain a better view of my readers and visitors.

Even frivolous questions get an answer. Serious questions are answered as completely as possible. The path for dollars to my pocket begins with a site visitor who subscribes to "STAT News." Once they decide I know a couple of things, and come to believe I can be trusted with their feelings, they may step forward and ask a question. A good reply generally creates a supporter, one who may also be a potential client. Paths to profits on your site may be quite different. But figure what they are, and enhance each step along the way as possible.

It's tough to do. About when you decide your visitors know nothing about the Web, a steady flow of them begin to point out authoritatively where you are screwing up.

What it comes down to is hunches and guest mates. But try to answer such questions as how old your average visitor is. Something of their economic status. And so forth. As mentioned, I take one input as representative of nine others not received.

Also try to answer questions relative to your business. Where are your visitors and subscribers on the Web cycle? What percentages are novices? Have a site? Want a better one? Your questions will be different than mine, but take the time to state them clearly. Then seek the best possible answers.

Behind all this, you have already defined your niche and target market. Disregard any input that is off target. Answer email, of course, but discount this input from your view of your visitors and subscribers. The objective is to continue to narrow your focus even further. Thus you are looking for input that will help you do so. In short, you are seeking a better definition of your perfect customer.

Write your newsletter and site pages targeted as closely as possible to your view of your typical reader and visitor. Write for the needs of your perfect customer. In time those who do not fit within your tight focus will unsubscribe and stop visiting your site. You will in fact have created an audience who for the most part are interested in your niche or focus.