I know there's a tool at Overture, but I'm not sure it's that useful? I put in photography, and most of the top results are nude photography, glamor, erotic, and so on. I've carved out a few niches, and I'm sure I could get more specific with the Overture tool ... but I'm hoping for some more general advice first.

Like I keep seeing search phrases I would have never imagined in my referrer logs, and I get a lot of clicks from a few queries where I'm on the third page. Since January, I get clicks for "Stret Dragon." "Star of Zion" comes up month after month, and it turns out "green chanels" is sending people my way. Even "happy little girl" ( I have some family portraits ). And then some of my articles come up for a dozen different ways of saying the same thing.

So far I've generally tried to be descriptive on my pages and in the handful of links I've been able to create myself. And I've tried not to use the same word or phrase too often and sound redundant. Haven't really put that much focus into targeting keywords. But Google is sending more and more people my way, and it seems like I ought to run with this momentum.

I know these are beginner questions, but would I increase my viewer ship by optimizing for queries that have been popular? Do I want to spend the time chasing phrases that seem connected to my site from Overture, or is it better to try to improve my rank for the queries people have already found me from? Given some of the weird ones, it feels like a guessing game.

Unless your website has a chance at being competitive for the main phrases such as photography, I would focus on creating larger amounts of varied content to try and capture the search tail.

Have you tried wordtracker?

Try to find some here free seo tools

Forget the overture "tool". On the useful scale it's just above chocolate teapot and elastic measuring tape.

Buy a day (or a week) at Wordtracker or a month at KeywordDiscovery.
Neither are exact tools but they will give you some grounding in what people are looking for.
And don't ignore your own site logs as the most valuable source of keywords. That's what you get found for already, with a bit of thought you can extrapolate more from these.

Hello. As far as I know Overture shows statictic for Yahoo PPC system. I think that we shouldn't trust it.

Used Overture (inventory overture of google) and Wordtracker

At first I was a big Wordtracker fan, but I've gone away from any of these tools lately. The reasons are:

1) The Internet userbase is becoming increasingly varied with each passing day.

2) Niche keywords are both easier to target and bring more qualified customers than their generic counterparts.

3) Neither tool has been especially accurate in terms of me finding keywords and phrases that I have and my clients have been found under. For example, the #1 phrase that I am found under for the SEFL site (kept private because quite frankly, I don't want people knowing about this little gem) shows no results for either WordTracker or Overture.

4) Niche keywords and phrases avoid the "all the eggs in one basket" scenario that quite often results from targeting a single keyword or phrase.

As Hirst said, the best way to find the keyphrases and keywords you want is to examine your stats and try to extrapolate others. It's something of a self-fulfilling prophecy in that regard, but it quite often works.

Originally Posted by ForrestCroce

Like I keep seeing search phrases I would have never imagined in my referrer logs, and I get a lot of clicks from a few queries where I'm on the third page. Since January, I get clicks for "Stret Dragon." "Star of Zion" comes up month after month, and it turns out "green chanels" is sending people my way. Even "happy little girl" ( I have some family portraits ). And then some of my articles come up for a dozen different ways of saying the same thing.

You're a pretty smart cookie, so I'm guessing you run a spell check. If you're coming up when people search for "stret" that can only mean someone liked you, and linked to you, but didn't know how to spell.

I would ignore your server logs, at least the part about search terms. I mean, those searches already work. People found your site because of them. Spend your time on words and phrases that people aren't finding you through.

I wouldn't ignore what is already bringing people to your site just because it's bringing in one or two people doesn't mean you can't optimize it to bring in more people.

Excuse me if I don't use the right terms but here are two examples:

I noticed one day that someone reached my blog looking for some information. I was curious about how they got there because I didn't have what they were looking for even though it's in my area. But then I decided to create a post on the subject using the words that brought that one person to my blog and created a post around it. Yesterday I noticed that I have had just over 100 visits to that entry in the past 30 days since I blogged about it.

The second example is a little similar. I noticed someone getting to my blog looking for something that I didn't have. So I did an entry about it on one of my blogs. Soon I noticed it was bringing in a lot of traffic to one blog. So I rewrote the entry for another blog keeping the same keywords (just changing content a little) and I just did a search for that term. I have the number two (Second blog entry) and number three (First Blog entry) spots on Google search for that search phrase. Also Below the #3 spot there is a space (indentation) and then there is yet another entry from the blog it first appeared on. And the feed of one of my blogs is the last entry on the first page.

So I have 4 impressions on the first page of Google for a search phrase that I wouldn't have had if I had ignored my stats from feedburner. The best part is that my two blogs top a major forum where this keyword is popular and above encarta that has an entry on this subject.

More on site logs;

Very often you will find a visitor from pages way back in the serps, result 100 say, so if one person has gone that far back there will be more that don't. Or maybe a page fitered up higher temporarily and got some clickthroughs, then dropped down again.
All these are real qualified visitors that found your page, maybe they didn't take any action, which could mean your page didn't meet their needs or expectations, so you can improve the position and the page for that phrase.

This is what genuine SEO is really about;
Getting views from real people rather than pointless rankings for some silly phrase.
Making pages that fulfil visitors needs, rather than what search engines "like".
Creating something that will survive in the ever changing world of Search Engines.

Then you can stop checking rankings every 1/2 hour, you won't be wondering if the latest update has dumped your site into oblivion and you won't be chasing link exchanges.

Okay, this was illuminating, so thanks to all who've lent their two cents! ( Or pennies, in Chris's case. ) I'm glad to hear the keyword suggestion tools aren't as crucial as they can sound ... every minute I spend at Overture is time I'm sitting at the computer and not improving my site. I wasn't very keen on doing a lot of keyword research, and now I can pretend there's a reason besides laziness.

Thanks for the explanation about how someone could get to my site when it's buried deep in the serps. I've been wondering a lot about that, because month after month I find things I don't score well for. I'm not sure how much I can do about it, but at least I don't have to wonder any more.

Last night "photomatix" turned up in my logs as part of a search phrase. It's nowhere on any of my pages, and I can't imagine it's in any anchor text that goes to my site. I'm not sure how that happened. But I've started writing an article on how to do "HDR" photography using Photoshop instead of having to buy an app like photomatix. Definitely some good came come out of paying attention to your logs.

Anybody know where the searches under Webmaster Tools come from?

It sounds like if you take some effort to write well for your site, and then give keywords only a small amount of thought, this isn't a bad approach?