• Register
• Log in

SEO Theory - SEO Theory and Analysis Blog

Algorithm analysis, Web community relationship analysis, SEO practices and techniques, industry news, etc.

• Home
• About SEO Theory
• SEO Glossary

Fundamental principles of keyword analysis

Posted by Michael Martinez on November 14, 2007 in Search Engine Optimization

Keyword analysis is one of the oldest, most fundamental tasks in search engine optimization. All that content enrichment and link building has to be driven toward the right keywords, the actual queries people use to find content on the Web.

There are four kinds of keywords:

1. Brands (high value)
2. Pop expressions
3. Seasonal descriptors
4. Random descriptors

“Pop expressions” includes anything mentioned in the news, in music, in movies, on television, etc. “Seasonal descriptors” are things like “Christmas lights”, “Easter baskets”, and “Valentine’s Day Flowers”. “Random descriptors” might be “bar mitzvah”, “winnie the pooh”, and “seo theory”.

A random descriptor is not very popular, not likely to become popular, and won’t generate much traffic. But you never know. A random descriptor may graduate to obtain either seasonal or pop value. It could also become a brand. In fact, you could say that random descriptors are the Harfoots of keyword types — they are the most common, the most “plain”, least conspicuous keyword types. Or you could say that random descriptors are the stem cells in the body of query spaces. You never know what they will become.

Search marketing principles tell us to chase the money, but does that mean you compete for the most active queries with 10s of thousands of daily views or does it mean you compete for more of the least active queries where competition is less fierce but traffic is down?

Keyword pundits have been pointing out for years that there is little actual monetization for 1-keyword queries. If you sell used cars you’re probably better off in the “city name used cars” query than in the “cars” or “used cars” query (unless you can ship cars across country and people are willing to pay for that kind of service).

Rsearch indicates that consumers do a lot of research online before making a purchase decision. Like it or not, the average Web-based business site needs to provide as much information as possible to draw in those price-unconscious consumers who just want to be reassured that they are about to make a good decision.

I’ve done plenty of research myself for personal needs. I’ve looked up replacement parts for computers and entertainment systems; I’ve looked up instructions on how to tune universal remote controllers to my TV, DvD player, and cable system (BTW — there is no easily findable, very useful information in that area); I’ve looked up information on car maintenance, cell phones, foods, restaurants, store locations, business numbers. You name it, I’ve done the research online and odds are pretty good that you have, too.

You cannot optimize for an informational query by throwing a price and shipping information in someone’s face. You have to understand the queries before you can really optimize for them. If I type in “price of AT&T cell phone”, I’m going to start looking for the lowest price I can find and then compare upwards, looking for additional features I may want. Other people may start at the top and compare downward, dropping features as the price goes down.

How do your Web sites handle such queries? It’s not the search engine that answers the query, it’s your content (or someone else’s) that really answers the query. Keyword research has to pick the best questions for the right answers you can give people. You can optimize a single page for 100 SEO questions or you can optimize a page for SEO, pizza, and Britney Spears.

However, you cannot optimize a page for everything. And in most cases I don’t think you really want to optimize 1 page for 100 queries. It’s a fun trick but in real business you want to make a good impression and help people find what they are looking for as fast as possible. Requiring people to scroll through 100 questions to find their answer is not user-friendly.

Your keyword analysis should start with the marketplace.

Can the service or product be found only online, mostly online, sometimes online, or not online at all?

Are there well-known provider brands or is the industry mostly small businesses? Is there widespread industry recognition or is this a niche or emerging market?

Are there firmly entrenched keywords or do the queries wander all over the place? By firmly entrenched, I mean do you have rock-solid search activity for a small number of queries or are the queries just randomly scattered across the stars?

Can the products or service be localized? Are people more likely to look in their own neighborhoods or do the products and services come mostly from one or a small number of specific communities?

Is the market young or mature? Is the market growing or shrinking?

Search activity can be a great indicator of where a market is in its life cycle. If you’re not looking at query trends (when the data is available) you’re missing most of the picture in keyword research. You need to know whether queries are cyclical, tied to news events, tied to entertainment activity, tied to brand value, etc. You need to know if interest in the products or services is growing or declining. For example, now is probably not the best time to optimize for “Harry Potter books” or “Lord of the Rings movies”.

But if you find you have missed the boat on a query trend, that doesn’t mean there is no reason to optimize for teh query. You may be able to breathe some new life into a dying market. You may be able to seize part of the shrinking pie from people who have become less attentive to the marketplace. Search activity may enjoy a very long, slow decline stretching out over several years.

Study the top-ranking sites

Pick some queries and look at who the top ranking sites are. Then go pick some more queries (similar to the first group) and look at the top ranking sites. Do you see the same sites over and over again? Even if there are only 2 or 3 such sites that tells you there is considerable competition in the query space (definition: A Query Space consists of all the queries and relevant content that pertain to a specific subject.).

The Theorem of Query Space Optimization tells us that queries remain productive only as long as there is search interest in them, and only as long as relevant content is promoted for those query spaces. Knowing who is optimizing for queries tells you something about the level of competition you’re facing. Some sites optimize for random queries through user-generated keyword traps. They are kind of cool. You go to a seemingly large content site and search on a keyword. The site has no content but generates a page based on the keyword for you.

Many a spammer has enjoyed seeing their keyword trap sites grow. So do many news agggregator sites like Topix. You can create a page for Michael Martinez SEO Theory that shows all sorts of advertising and Google will index it for you. It’s that simple. Now if you’re not sure where you found that randomly generated content somewhere down the road there is a chance you’ll find it in a search engine.

Topix grabs some of my name space, gets another page for internal link boosting, and increases its search visibility all because I linked to a perfectly worthless page that neither you nor I could care less about. But if enough people search Topix for “michael martinez seo theory”, they may just notice and set up a permanent page. Their search tool will point you to their permanent content pages if there is a good match.

Keyword traps have been working as useful doorways for entertainment news for years. Google loves that kind of worthless content. But it’s worthless content that provides the Web operator with very valuable data: you find out what people are searching for and what people are willing to link to.

Well, we can’t all be Topix and earn Google love for worthless, spammy ad-laden pages derived from user queries. But there are other ways to gather data and to do it more quickly. Google and Yahoo!, for example, will show you what people have been searching for if you use their advertising tools. And Google Trends helps us look at queries with a LOT of space.

Even there, however, you have some limits. But query activity is not limited to the major search engines. If you run a forum or blog-hosting service and can capture query data, people will show you what they are interested in. If you have a large content site and provide site search, you may be able to see the queries and learn what people’s interests are. Comparing query data from month to month tells you where trends form and where cyclical patterns develop.

And as many keyword pundits have said through the years, just looking at your own referral data from your Web server logs tells you a great deal. Of course, this is all good to have if you have it, but when you’re just looking for a new market or have not yet created a Web site you’re at a loss for data collection opportunities.

Not to worry, the competitors will tell you what you need to know. Just don’t waste your time looking at their keywords meta tags.

Evaluate what the competition values.

Although most Web sites probably do put honest data into their keywords meta tags, there is no law saying they have to. Only Ask and Yahoo! even care about the keywords meta tag any more, and it’s not above some people to put misleading keywords into that tag. I would never start competitive analysis by looking at the keywords meta tag. You want to validate what you find there in a competitor’s Web site, and he cannot help but show you what he is competing for elsewhere.

The first place to look for keywords is in the title tag and page URL. Most people cannot resist the urge to put their beloved keywords into titles and URLs. Refraining from using keywords in those locations gives you a competitive advantage, but if you don’t know how to optimize without putting keywords into titles and page URLs you’re at a competitive DISadvantage.

If the competition is using a cookie-cutter template or is simply following cheap SEO advice they bought off the Web somewhere, they have pretty much left their playbook open on a counter top for anyone who knows what to look for to read. You can tell which blogs influence the designers by whether they use dashes ( - ), bars ( | ), or other special characters in them. You can tell who their SEO idols are by whether they brand company name into the title or not, put the company brand first or last, or omit the company name altogether.

You can tell whether someone has optimized their site by comparing titles from multiple pages. You can tell if they are relying on cheap SEO tools like link checkers and the Google Toolbar or if they really know something about search engine optimization by comparing their title tags to query research tools. If you don’t put a full query into your title tag, you’re either going for a really long query or you bought the wrong eBook.

One of the most popular queries for people who click through to SEO Theory is external links versus internal links. It’s no accident that I chose an appropriate expression for the title — I know the words would be included in the URL. I included the query in the title tag. I didn’t just brand the post with some crazy, unrelated title like “SEO tips you have to use”.

When you’re done looking at the page title and URL, look at the meta description tag. If there is no tag either the site is using a cheesy template (like SEO Theory does — and I appreciate all helpful hints but please don’t tell me there is a plug-in — I leave this blog plain and ugly for a reason) or else someone didn’t know the value of using meta descriptions. It could also be the optimizer is extremely confident in his/her ability to ensure that search engines find the most relevant text on the page to show in snippets.

When you’re done looking at the meta description, look at the on-site navigation anchor text and its placement. If the real value-passing anchor text is in the open either the Web designer feels he has nothing to hide or else he just doesn’t realize what he is telling you. Me, I don’t care. SEO Theory has an SEO Glossary page. I probably need to update it, now I think about it.

How many query expressions do you actually see on the visible page? If you don’t see any, someone ain’t optimizing for the queries. But if you see a match between the keywords meta tag and everything else then you can probably trust the keywords meta tag to tell you which queries are most important to the site optimizer.

But the point is that you don’t need to look at someone’s keywords meta tag to understand which queries are important to them. It’s nice when the competition leaves the playbook laying open, unwatched, but you can glean the information you need from other sources.

Just knowing what people optimize for gives you a starting place. You can go look at current search traffic. You can go look at archived images of their site and see how long they’ve been optimizing for those words. You can look at their inbound links and see what anchor text they bought, begged, or stole. You can look at every mention of the Web site and see if they promote the site or promote the site for keywords.

Knowing how the other guy promotes his site tells you more about what he wants to rank for than about how well his efforts are being rewarded by search engines. Most SEOs (including me) get more rankings by luck than by effort. There are relatively few SEOs (if any) who get precisely what they are looking for and nothing else. Maybe the very best black hat spammers and cloakers can do that kind of precision SEO. I’ve never seen anyone do it.

So, I have to wrap this up for now but I’ll come back with more on keyword analysis. There is a lot left to say on the subject.

Don’t believe me? Just run a few queries for “keyword analysis”.