Site Architecture

Presented by Stoney deGeyter

Why should we care about site architecture?

Good Site Architecture

  • Helps SEs find important pages
  • Helps users find them
  • Helps SEs determine page relevance
  • Helps users & SEs have to think less

Bad Site Architecture

  • Roadblocks SE spiders
  • Decreases rankings
  • Confuses visitors
  • Reduces sales

You want your site to be a "no thinking zone" where neither SEs nor human visitors have to try to figure out what to do next.

Site Architecture: foundation for good SEO

Elements of Site Architecture

  • Domain and URL structure
  • Link & navigation structure
  • Document/page structure

Duplicate content issues

Duplicate content can plague you like a virus (or two). When two pages are identical, one gets booted from index. When SE spiders find too many duplicate pages, they may not spider your entire site. They assume the whole site is made up of duplicates, even when unspidered pages may be unique.

Link Structure

How are you linking to pages within your site navigation (with WWW or without)? You want to be consistent, no matter which way you link.

Absolute vs. relative links

  • Absolute = full URL, always the same no matter where the link is
  • Relative = based on location where the link is

Positives and negatives in both. Stoney's preference: go with absolute. You know exactly where SEs and visitors are going to go, no room for misinterpretation.

When linking on secure pages, always use absolute to make sure you stay in secure when you need to, and leave secure when you need to. (i.e. links should include either HTTPS or HTTP to make sure visitors go where you want them to). Otherwise, can lead to duplicate content issues if secure pages get indexed when you want non-secure version indexed (or vice versa).

Linking to home page: don't link to "index.html" -- link to alone without "index.html" designation. At first, SEs will think and are two pages (duplicate content). Eventually SEs will figure it out, but in the meantime, you're splitting your link pop.

Search friendly links

Keywords in URLs play a very minor role. But if you can get them in, do so. Value in having "SE friendly links" with keywords. Also helps humans figure out where the link is going, so they can decide if they want to go there or not.

Bad links: links written by javascript won't be followed by SEs, so they're not spiderable. Avoid them unless you're trying to block the spiders for some reason.

So-so links: SEs may or may not be able to spider them. Don't take a chance, don't make SEs have to think. Don't leave anything to chance - so fix these when you can.

Good linking practices:

  • Avoid Flash and Javascript whenever possible
  • Keep links in control (less than 100 per page)
  • Every relevant page needs at least one incoming link (no orphan pages)

If you have pages with similar content (for example: product pages where same page is in two categories with two different URLs) and you link to both pages, you're splitting the link flow between those two pages, decreasing the value of both pages.

Try to set it up so each piece of unique content is available through one unique URL only, so all links can point to that one page.

Site navigation: purpose is to get people to where they want to go. Header image should be linked to home page. Top and side navigation is common. Make sure your main navigation links to all the main category areas on your site. This indicates to the SEs what is the "top layer" of your site.

Usually not a good idea to link to every page from your home page, unless your site is under 20-30 pages max. That's what the site map is for. Link to important pages, categorize to help people quickly and easily figure out where to go.

Footer navigation: good place for SEO related links, admin/contact/company related pages. Try to avoid the big block of link text at the bottom of the page. Looks ugly and not usable.

Goals of good site navigation: intuitive and non-complex.

Breadcrumb navigation: helps people know where they are and helps them navigate back to previous sections without having to figure out where those sections are in the overall "big" navigation. A lot of people don't actually use the breadcrumb navigation, but they're good for visual cues.

Link anchor text

"Click here" and "read more" are terrible link anchor text. Make anchor text scannable and informative about where the link is going to go. Include keywords.

Links should accurately reflect the information on the destination page. Nothing wrong with having long hyperlinked sentences (just don't go overboard).

If you're using two links to the same place, make the first link descriptive anchor text. May be OK if second link is "click here" because it appears SEs may be paying more attention to first link anchor text.

Cross link related products, either by straight HTML or by database links in your back end. Helps both users and SEs categorize the page and find related content.

Product categorization can create duplicate content issues by making same products available through two different URLs when they're included in two categories. Solution: have a "master" category for each product. No matter how customers get to the product, the final product description page must be fed from master category/product URL.

When you have multiple products on the same page, allow some kind of filtering so customers can narrow down the choices. Eliminates long string of "next" pages that might not be so good for users or SEs.

Question: what is the effect on the SEs of changing your site navigation?

Answer: when you change navigation, you will see some temporary dropoff. You have to decide if the long term payoff is worth the short term pain. Putting keywords in URLs is not a good enough reason. Fixing duplicate content issues might be.

Keep shopping cart pages hidden from SEs. No reason for them to index these pages. Keep them out, whatever you've got to do. You don't want search engines adding products to a shopping cart.

SEs can't index links that are blocked by a form. In some rare instances, they're trying to fill out forms with some larger sites, but why take the chance? Make sure your product/content pages are accessible without going through a form.

Sitemap: unless you're a very small site and every page links to every other page. Good place for sitemap link is in the footer. You may need master sitemap and sub-maps if your site has thousands and thousands of pages. Make it useful for humans as well as SEs. Don't rely on sitemap to get visitors to those pages, but it can be helpful.

Exclusion: robots.txt can be used to exclude SEs from certain pages. Be careful -- can accidentally exclude spiders from your entire site. You can use this to exclude the policy pages, for instance. There are different strategies, might want to read up on this further.

Nofollow attribute: originally used to indicate a link to a site you weren't willing to vouch for. Now, google says nofollow will avoid passing "link juice." Stoney recommends using nofollow on "click here" links to help direct SEs to more valuable anchor text links. Preferable to use your site architecture itself to "sculpt" PageRank, but can still use nofollow on links to pages like contact and policy pages.

Noindex/Nofollow META tag: prevents indiviudal pages from being indexed and/or links from being followed.

On a monthly basis, run a broken link check. Broken links happen all the time, and it's a good thing to check. You don't want one simple change on your site to prevent SEs from finding pages. He recommends Xenu Link Sleuth (free download) to check links.

If you change a URL or delete an old page, be sure to redirect to newer, more relevant page. 301 permanent redirect is usually the best/safest way to go, to make sure you pass along traffic and link juice. Leave the 301 there for as long as you can. No reason to get rid of them, because they'll catch traffic from those old bookmarks and links.

Site hierarchy: lay out your site in a way that makes sense to SEs and to people. Stony recommends structure that's "amazingly similar to directory structure" where there's a home page, leading to main topic pages, each of which leads to subtopic pages, each of which leads to more detailed pages, etc. You want the organization to make sense, with each unique section of the site to be pretty much self contained so pages that are relevant to a specific topic, they link only to that topic.

(It is OK to cross link between sections when the pages are related.)

Title tags: should be unique for each page. Best to build this in from the get-go without having to go back and "fix" the database. OK to have common elements in different title tags (company name/brand, for instance) but each page should have some unique element. Useful for branding to have business name or brand in all titles.

Put business name at the front of the title if it makes sense (brand building or already well known brand), or at the end if that's where it works best, or leave it out entirely if the brand isn't relevant or significant.

Description META tag: not every page needs a META description. Use for section-level pages. Don't need it for article pages targeting long-tail terms because you want SEs to find the relevant snippet from the text on the page. Use when you're targeting a few higher-competition phrases for the page to help control the snippet that shows up.

Keyword META tag: not even worth discussing.

Content: create unique for each pages. Don't rely on default product descriptions. (You don't want to be the same as the manufacturer and your competitors.)

Use textual interlinking to connect related content on your site whenever possible. Helps reinforce relevance and makes for better user experience.

On-page content:

  • Clearly written
  • Accurate
  • Research and use keywords
  • Use image alt attributes

If you have video or audio content, transcribe it. Good for users who can't see video/hear audio, helps SEs index the content. Produces a lot of good content the SEs otherwise wouldn't find. Audience member: teach Dragon Naturally Speaking the words in your audio and it will transcribe for you, then all you need to do is proofread.

Hx Heirarchy:

Lay out like an outline you used to set up for term papers when you were in school.
One H1 at the top for the overall subject of the page
Multiple H2s, one for each main topic
Multiple H3s, one for each subtopic, grouped under appropriate main topic
And so forth...
Don't make every heading an H1. Don't put H1 down below H2.

Eliminate code bloat:

  • Excessive tables -- tables are intended for tabular data, not formatting. If you use them for formatting, simpify your table structure as much as possible, and keep in mind how they're read by the SEs: they read cell-by-cell, starting at upper left corner, row by row.
  • On-page Javascript and CSS: eliminating makes pages load faster
  • HTML formatting (like font tags)

General hints

Use ALT attributes for images, especially those that are linked.

Use valid HTML whenever possible. Will not improve rankings, but may help SE spiders get through your page successfully.

Sight-impaired accessibility is also important, and becoming more important over time. Keep in mind needs of screen readers, provide adjustable fonts.

View your website in a text-only browser to make sure it can be read properly to make sure information that needs to be found can be found.

View your site with CSS turned off to make sure it's usable for CSS-less browsers.

Check your site on a mobile browser. More and more people are using their cellphones for web access.

Create printer friendly pages (you can use CSS for this rather than creating whole new pages). Test what your printer friendly pages look like.

Make sure you don't require cookies for standard browsing. People get nasty-looking warnings if they have cookies turned off (and SE spiders don't accept cookies).

Site architecture should be a priority in your SEM campaign. Doesn't do any good to have great pages with great content if the SE spiders can't get to it and index it. A search friendly structure will help achieve stronger SEO results: improved spidering/indexing, better customer usability.


How long should META description be?

Shouldn't be longer than content on the page. Keep it simple. Cover what you want people to see about the content on the page (snippet that shows in the SEs)

Is it a problem to have hyphens in domain names?

Avoid hyphens. It's hard to tell people how to get to your site when you have to "read" hyphens out loud.

September 22, 2008

Learn more about the ways Diane can help improve the performance and profitability of your business web site, or request a no-obligation personal consultation, by visiting


Great.but this article's structure confused me. why don't you classify your content by three Elements of Site Architecture?

Evidently I should have made it more clear -- this is a transcript of a session that took place recently at the Small Business Marketing Unleashed conference in Columbus, Ohio. I wrote up the presentation, but the content and structure of the presentation was up to the presenter, Stoney DeGeyter.

If you're really curious why Stoney didn't organize his presentation according to the "three elements of site architecture" (whatever those might be), I guess you'd have to ask him. :-)

Comments closed after 30 days to combat spam.

Search Engine Guide > Diane Aull > Site Architecture - SBM Unleashed