Skip to content

5 search practices you need to improve, now

June 11, 2010

I was blown away by a recent blog post by SEOmoz, which describes various correlations between page attributes and ranking in Bing and Google. They tested thousands of existing pages for the top AdWords categories and cranked out one of the most rigorous analyses of actual ranking factors I have ever seen. Two things struck me most about the analysis:

  • I’m continually amazed at how hard the search engines work to thwart attempts to rig natural search results.
  • Search strategy needs to be cued into the continual changes the engines are making. That doesn’t mean trying to find clever ways of hacking the system. It means finding ways that you don’t unwittingly hurt your own efforts by being out of date in your guidance.

I won’t go over the report in much detail. You can read it. It’s already concise. And the most important factors relate to links, over which you have limited control. But I do want to point out five web publishing practices that are fairly pervasive in the industry, for which you probably could do a better job. If you do nothing else with the SEOmoz post, improve these five web publishing practices.

1. Don’t worry about keywords in the title tags or H1 tags. The study shows a negative correlation with having keywords in title tags and H1 tags for both Bing and Google. That means you actually get knocked down the rankings for doing what most web publishers try to do above all else. This is kind of like what happened when metadata stopped having any impact on the search rankings. In this case, it not only doesn’t have much effect on the rankings, it has a negative effect. The study concludes “Everyone seems to be optimizing their title tags these days (appeared in Google: 11,115 vs. Bing: 11,143). Differentiating here is hard.” So it’s not like you shouldn’t put keywords in the title or H1, but it won’t get you much.

2. Take the time to get the URL right. Google and Bing don’t agree on this one. But Google gives big benefits for having keywords in the URL. The closer the keywords are to the root directory, the more advantages the algorithm gives you. The best possible case is to have the keywords between www and .org or whatever top-level domain you have. Otherwise, try to get the keywords as close to the first / as you can. This is a switch for a lot of web publishers who have legacy environments with clear URL structures. For these folks, it is very difficult to add new URLs and meet this requirement. This puts a lot of pressure on the webmaster team to create and manage new high-level domains. But it is usually worth the effort.

3. Focus on the body copy. A lot of teams do this already, but some think they are done when they get the title, H1 and H2s right. Google weights having keywords in the body copy higher than Bing. For Google anyway, you can’t afford not to have at least a couple of instances of the keyword or a stemmed form of it in the body. The study did qualify this however: It’s not about the number of times a keyword is in the body, it’s about the keyword density. One instance in a short, pithy page is better than two instances in a page that makes users scroll.

4. Focus on alt attributes. Alt attributes are pieces of text embedded in image code that allow screen readers to tell sight-impaired users about the image. Though primarily used for accessibility reasons, both Google and Bing give you a lot of credit for having keywords in alt attributes. This makes sense because they can’t see whether images are relevant to the query or not. All they have to go on are alt attributes.

One note of caution, however. Please don’t spam alt attributes. The alt attribute should be natural language that makes sense when read by a screen reader. If you can’t fit keywords into all your alt attributes without making it sound forced, don’t sweat it. Just don’t shoehorn them in to pump up your ranking artificially. If enough sites start doing that, Google and Bing will change their algorithms to thwart those attempts. So for the sake of SEO practitioners everywhere, please be honest and transparent in your use of alt attributes. We don’t want another situation like we had with metadata, title and H1 tags on our hands.

5. If you have a choice, don’t choose .com domains. In their tests, .org and .gov domains did much better than .com domains, all things considered. So if you’re considering starting a new site, use .org if you can get it. By all means buy .com, but don’t automatically use it as the default.

I wrote a post when we first started this blog about how dumbfounded I was by a particular high profile page in our environment ranking poorly despite our best efforts. Those familiar with that post will see the five recommendations I list above as familiar. In short, I used our best guidance, thinking I was doing all the right things, but four of the five things we did were the wrong things.

  • We were not able to get much keyword density in the body copy  because it was a speech by IBM Chairman, CEO and President Sam Palmisano, and we weren’t about to change his words. The best we could do was to write a preface before the speech transcript with one instance of the keyword in it. In a half-hour speech, that is not nearly enough keyword density for Google to give us credit for having keywords in the body copy.
  • Our URL string was too long to get much benefit from the keywords we built into it.
  • SEOmoz is the first study I’ve seen that confirms my hypotheses about competition from .org and .gov domains, which brings a bitter sweet sentiment. If I had known how hard it is to compete with those domains, I would have chosen different keywords.
  • And the title and H1 issues shocked me. We worked hard to get the keywords in there, and it only hurt us.
  • The only thing we did right was to focus on alt attributes. We worked keywords into them where appropriate and probably got some benefit from it. But it was balanced out by the four things we did wrong.

Perhaps the biggest thing I’ve learned in all this is never to take your guidance as gospel, but humbly change your sacred guidance when research indicates it no longer works. Also, you can never do enough research. That reminds me, I’ve got work to do to change our guidance, including in our book, where we advise readers to put the keyword in the title and H1 tags. It doesn’t mean our guidance is wrong, it just might not be as important as it seemed before this study.

7 Comments leave one →
  1. Dey Alexander permalink
    June 13, 2010 2:03 am

    It’s interesting that reading the SEOmoz article, I didn’t end up with the same takeaways that you did. But thanks for the pointer to it.

  2. Cliff Tyllick permalink
    June 13, 2010 3:18 am

    I think that, especially in advising people not to use .com domains if they can avoid it, you are making the mistake of assuming that correlation means causation.

    The study showed that .com domains performed less well than .org domains in the search rankings. It doesn’t mean that the type of domain is itself the cause of that difference. It could be that the factors that impair SEO tend to occur more frequently on e-commerce and other .com sites and less frequently on .org or .gov sites. In fact, I would argue that this differential frequency of other factors is far more likely to be the reason for this observed difference in performance.

    And it could well be that many of “the” factors that impair SEO were not recognized in the SEOmoz study. So even if the folks at SEOmoz could correct for the impact of the factors they did recognize, they couldn’t correct for all causative factors.

    After all, it’s not as if every .com url has .org and .gov clones we can use to directly measure the effect those three letters have on SEO.

    • June 13, 2010 5:14 pm

      I get your point. And we at writingfordigital don’t all agree on my somewhat strong conclusions. So thank you for bringing your point of view out here.

      In my defense, I did say, “all things considered”, when advising about .com domains. In other words, by itself, a .com domain will not hinder you enough to be a deciding factor, but all things considered, .com domains will not perform as well as .org domains. In general, none of these factors is a silver bullet. And there are more than 200 factors that go into SEO performance. But if you have two identical pages and one is and the other, will tend to perform better. Rand himself draws much the same conclusion in his article.

      The reason I focused on domains among the two dozen factors Rand brings up is we have seen how difficult it is to compete against .org and .gov domains in practice. Our pages do not rank ahead of comparable pages with .org and .gov domains, even if we have a lot of quality link referrals and we have optimized to the best of our ability, often better than the competition with domains. That’s what the case study I link to is about. So it was interesting to see statistical validation of our anecdotal evidence.

      I get the thing about causation and correlation, believe me. But all we have to go on is correlation. That’s what statistical evidence give us. Statistical evidence does not give you cause and effect, but trends and tendencies. If that is all you have to go on, you act on it. It is better than flying blind. If you have a large enough sample size to gain statistical significance, and your data is outside the margin for error, you can use the evidence to make intelligent decisions about your content. Rand even says he has such a huge pool of data, the margin for error is negligible. So even a small deviation from 0 (p0sitive or negative) is significant. It doesn’t mean you will see the results he sees in his tests, but you will be more likely to, all things considered. That’s good enough for me.

  3. June 14, 2010 7:43 am

    The SEOmoz study doesn’t suggest to abandon the .com domain in favor of a .gov (not that you would have a choice unless you happen to be a government agency, which should already have a .gov domain) or a .org domain.

    .org domains are often used for non-profits, including professional organizations like and that publish lots of quality content and gain links from other credible sites, and Wikipedia too which generally scores well, so the positive correlation between SERP ranking and the .org domain doesn’t come as a surprise, but the domain is hardly the cause for the good ranking (and the study even states that Google’s claims not to give priority treatment to certain domains appears true).

    ccTLDs are a different story; they can positively influence local search results but anecdotal evidence suggests moving a site to a ccTLD can equally do harm if the site is mostly linked from international pages. Those links seem to be considered less relevant, lowering the ranking of the pages in search results.

    • June 14, 2010 12:42 pm

      Good points Klaus,

      For the record, I didn’t suggest that either. Just that, if you’re starting a new site and you have a choice, to consider .org over .com. That might not even be appropriate if the site’s primary purpose is commercial. But for an information site, like, .org looks like the better choice. As I said, Rand suggests that himself. This is his exact quote

      “# The .org TLD extension is surprising – do these sites earn more links? Do they have less spam? Perhaps they tend to be less commercial and have an easier time garnering references? In any case, we’re happy to be!
      # Don’t forget about the exact match data from above – .com is still probably a very good thing (at least own it if you’re using a different extension)”

      So he does qualify it for all kinds of other possible factors. And you’re right, .org domains tend to get better link equity than .com domains because they tend to be impartial, whereas commercial sites tend to be focused on a company’s offerings. Painting with a broad brush here. So it might not be a question of the algorithm as much as the landscape of .org and .com domains, as you suggest.

      One thing we can conclude, it does seem to be more difficult to compete with .org domains than .com domains, whatever the reason. If a keyword has a lot of .org, .gov and .edu competition, I recommend finding a different keyword, unless it is a phrase that is central to your mission.

  4. June 14, 2010 12:52 pm

    Another comment worth reading about the SEOmoz’ study: SEO Is Mostly Quack Science

    I wouldn’t quite agree with that conclusion but SEO does indeed have science, art and randomness components🙂


  1. 2 SEO overstatements from my previous post « Writing For Digital

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: