On-Site Search Engine Optimization
If you’re in SEO, you probably hear this question a lot. Sadly, there’s no cut and dry answer, but there are sets of best practices we can draw from and sharpen to help get close. In this blog post, I’m going to share our top recommendations for achieving on-page, keyword-targeting “perfection,” or, at least, close to it. Some of these are backed by data points, correlation studies and extensive testing while others are simply gut-feelings based on experience. As with all things SEO, we recommend constant testing and refinement, though this knowledge can help you kick-start the process.
HTML Head Tags
Title – the most important of on-page keyword elements, the page title should preferably employ the keyword term/phrase as the first word(s). Using the keyword term/phrase as the very first words in the page title has the highest correlation with high rankings, and subsequent positions correlate nearly flawlessly to lower rankings.
- Meta Description – although not used for “rankings” by any of the major engines, the meta description is an important place to use the target term/phrase due to the “bolding” that occurs in the visual snippet of the search results. Usage has also been shown to help boost click-through rate, thus increasing the traffic derived from any ranking position.
- Meta Keywords – Yahoo! is unique among the search engines in recording and utilizing the meta keyword tag for discovery, though not technically for rankings. However, with Microsoft’s Bing set to take over Yahoo! Search, the last remaining reason to employ the tag is now gone. That, combined with the danger of using keywords there for competitive research means that at SEOmoz, we never recommend employing the tag.
- Meta Robots – although not necessary, this tag should be sure NOT to contain any directives that could potentially disallow access by the engines.
- Rel=”Canonical” – the larger and more complex a site (and the larger/more complex the organization working on it), the more we advise employing the canonical URL tag to prevent any potential duplicates or unintentional, appended URL strings from creating a problem for the engines and splitting up potential link juice.
- Other Meta Tags – meta tags like those offered by the DCMI or FGDC seem compelling, but currently provide no benefit for SEO with the major engines and thus, add unnecessary complexity and download time.
- Length – Shorter URLs appear to perform better in the search results and are more likely to be copied/pasted by other sites, shared and linked-to.
- Keyword Location – The closer the targeted keyword(s) are to the domain name, the better. Thus, site.com/keyword outperforms site.com/folder/subfolder/keyword and is the most recommended method of optimization (though this is certainly not a massive rankings benefit)
- Subdomains vs. Pages – Despite the slight URL benefit that subdomains keyword usage has over subfolders or pages, the engines’ link popularity assignment algorithms tilt the balance in favor of subfolders/pages rather than subdomains.
- Word Separators – Hyphens are still the king of keyword separators in URLs, and despite promises that underscores will be given equal credit, the inconsistency with other methods make the hyphen a clear choice. NOTE: This should not apply to root domain names, where separating words with hyphens is almost never recommended.
- Number of Keyword Repetitions – It’s impossible to pinpoint the exact, optimal number of times to employ a keyword term/phrase on the page, but this simple rule has served us well for a long time – “2-3X on short pages, 4-6X on longer ones and never more than makes sense in the context of the copy.” The added benefit of another instance of a term is so miniscule that it seems unwise to ever be aggressive with this metric.
- Keyword Density – A complete myth as an algorithmic component, keyword density nonetheless pervades even very sharp SEO minds. While it’s true that more usage of a keyword term/phrase can potentially improve targeting/ranking, there’s no doubt that keyword density has never been the formula by which this relevance was measured.
- Keyword Usage Variations – Long suspected to influence search engine rankings (though never studied in a depth of detail that’s convincing to me), the theory that varied keyword usage throughout a page can help with content optimization and optimization nevertheless is worth a small amount of effort. We recommend employing at least one or two variations of a term and potentially splitting up keyword phrases and using them in body copy as well or instead.
- H1 Headline – The H1 tag has long been thought to have great importance in on-page optimization. Recent correlation data from our studies, however, has shown that it has a very low correlation with high rankings (close to zero, in fact). While this is compelling evidence, correlation is not causation and for semantic and SEO reasons, we still advise proper use of the H1 tag as the headline of the page and, preferrably, employment of the targeted keyword term/phrase.
- H2/H3/H4/Hx – Even lower in importance than the H1, our recommendation is to apply only if required. These tags appears to carry little to no SEO value.
- Alt Attribute – Surprisingly, the alt attribute, long thought to carry little SEO weight, was shown to have quite a robust correlation with high rankings in our studies. Thus, we strongly advise the use of a graphic image/photo/illustration on important keyword-targeted pages with the term/phrase employed in the alt attribute of the img tag.
- Image Filename – Since image traffic can be a substantive source of visits and image filenames appear to be valuable for this as well as natural web search, we suggest using the keyword term/phrase as the name of the image file employed on the page.
- Bold/Strong – Using a keyword in bold/strong appears to carry a very, very tiny amount of SEO weight, and thus it’s suggested as a best practice to use the targeted term/phrase at least once in bold, though a very minor one.
- Italtic/Emphasized – Surprisingly, italic/emphasized text appears to have a similar to slightly higher correlation with high rankings than bold/strong and thus, we suggest its use on the targeted keyword term/phrase in the text.
- Internal Link Anchors – No testing has yet found that internal anchors are picked up/counted by the engines.
- HTML Comments – As above, it appears the engines ignore text in comments.
Internal Links & Location in Site Architecture
- Click-Depth – Our general recommendation is that the more competitive and challenging a keyword term/phrase is to rank for, the higher it should be in a site’s internal architecture (and thus, the fewer clicks from the home page it should take to reach that URL).
- Number/Percentage of Internal Links – More linked-to pages tend to higher rankings and thus, for competitive terms, it may help to link to these pages from a greater number/percentage of pages on a site.
- Links in Content vs. Permanent Navigation – It appears that Google and the other engines are doing more to recognize location on the page as an element of link consideration. Thus, employing links to pages in the Wikipedia-style (in the body content of a piece) rather than in permanent navigation may potentially provide some benefit. Don’t forget, however, that Google only counts the first link to a page that they see in the HTML
- Link Location in Sidebars & Footers – Recent patent applications, search papers and experience from inside SEOmoz and many practitioners externally suggests that Google may be strongly discounting links placed in the footer, and, to a lesser degree, in the sidebar(s) of pages. Thus, if you’re employing a link in permanent navigation, it may pay to use the top navigation (above the content) for SEO purposes.
- Keyword Location – We advise that important keywords should, preferably, be featured in the first few words (50-100, but hopefully even sooner) of a page’s text content. The engines do appear to have some preference for pages that employ keywords sooner, rather than later, in the text.
- Content Structure – Some practitioners swear by the use of particular content formats (introduction, body, examples, conclusion OR the journalistic style of narrative, data, conclusion, parable) for SEO, but we haven’t seen any formal data suggesting these are valuable for higher rankings and thus feel that whatever works best for the content and the visitors is likely ideal.