| 6:06 am on Jan 18, 2008 (gmt 0)|
The first thing is to understand that H tags are NOT style information - they are semantic informtaion. Yes, browsers do attach some default style for rendering each one. But you can, and should, establish your prefered style for each H tag via CSS.
So H tags are intended to communicate semantic information - they set forth the document's structure, much as an outline would:
H1 - the document's main topic
H2 - a sub-topic
H3 - a sub-sub-topic
H2 - a second sub-topic
|all have their introduction in H3 |
If by "introduction" you mean a summary set of sentences describing the page, then this is not an intended use of any H tag. Any H tag should contain a heading - a short description of the following section of the document, most commonly displayed in one visual line.
Now, on to the question of sending the best relevance signals to Google. The entire set of H tags, and especially the H1, have historically been abused on a wide-scale by both webmasters and software makers. There have been stretches of time where Google could not use H1 as a signal of any kind, and trated it only like any other on-page text.
Today we seem to be in a space where at least the H1 and H2 are again useful for discovering out relevance signals on many pages where they are in use. H tags are a best practice - but they are not required and a page can rank well without them.
An H tag does not automatically press a magic algorithm button that says "rank these keywords highest." The situation today is more subtle and nuanced than that. Instead, the signal that H tags give reinforcse other signals, both on-page and on-page, and helps to clarify the page's topic. Or if mis-used, it might conflict with other signals and actually blur the picture.
This all goes back to the roots of HTML itself, where the "M" stands for "mark-up". The assumption is that we begin with a document and then "mark it up" to communicate a richer sense of its semantic information and structure.
| 7:12 am on Jan 18, 2008 (gmt 0)|
A very intelligent reply. I learned about outlines in 9th grade (back then).
If you're learning how to make your website better, you'd do well to listen to tedsters advice.
Increasing to H1 could help as long as you don't change it's intended purpose (i.e using .css to alter format to blend with other H tag references).
Short and sweet - not a sentence.
Sub-highlights could be used for structure and clarification.
| 7:38 am on Jan 18, 2008 (gmt 0)|
Wow.. I love the webmasterworld Forums..
Thx Guys.. Very informative answers
| 9:31 pm on Jan 18, 2008 (gmt 0)|
A heading is simply a short piece of text that sits above, and introduces, the following paragraph, list, table, or form.
The W3C HTML Validator contains a tool to show if you are using headings in the right way. It lists them as bullet points. If that list is not a summary of your document, then you are not doing it right.
| 5:37 am on Jan 21, 2008 (gmt 0)|
|Presently my entire sites pages all have their introduction in H3. Will I get more attention from Google if I change them to H1. |
Page position may be as important as H tag. The first words of a page, some believe, are most important to Google (regardless of the tag). The same webmasters seem to believe the same weight-position theory with text applies to page title tags.
<title>Important Words Here Blah Blah Blah</title>
Tedster, what do you believe? Are title tags weighted by word position or diluted by title length (density)? I don't know that I'm convinced of the theory, because there doesn't seem to be great logic behind it.
What about text and link position on a page? Is the stuff nearest to the top given priority rating/ranking? That one makes sense.
| 6:47 am on Jan 21, 2008 (gmt 0)|
What is obvious is that words near the top of the file often get used for the snippet when there is no meta description.
Now, does that mean that near-the-top also gets extra weight from the ranking algo? I think it used to, very much, but today, not so much and possibly not at all. Today's algo is more complex in many ways, and Google's snippet team functions separately from the "relevance" team that works on the ranking algo.
For example, Google (and the other majors SEs as well) can analyze a page in terms of content blocks. So even though the snippet team may plug the top menu choices into the snippet (wish they would stop that) the relevance algo can pull out just the content area and look at that "unit" separate from the navigation, the footer, and so on.
So within the content block alone, does near-the-top get some extra juice? Hard to say. There can be all kinds of "pre-head" text that really doesn't mean much to the search algo - branding and marketing blah-blah-blah and that kind of thing. I'm thinking that an early line of text - one that reinforces or partially echos the title element - can help out. But other kinds of text in the same spot will not get a boost just because of its position in the source code.
So I'd say it's more of a holistic picture today, especially when you use a well structured page template with clean semantic mark-up. If your View Source is more like code soup, and if it varies significantly from page to page, then you are making Google's job harder. In that case those early lines of text just "might" still get a little extra juice. But the whole site may be getting so little juuice overall that it barely matters.
Part of my thinking here also comes from noticing that source-ordered content (using CSS positioning) doesn't seem to give the same edge that it used to. I can't say I've explicitly tested this, it's just the overall sense I get from working with a lot of pages and the rankings of a lot of sites.
[edited by: tedster at 8:11 am (utc) on Jan. 21, 2008]
| 6:59 am on Jan 21, 2008 (gmt 0)|
I missed one --
|Are title tags weighted by word position or diluted by title length (density) |
Not so much anymore for this one, either. In fact, it just might be one of those great holdover ideas from the '90s that live on now as myth.
Now before the flames start, let me share a recent happening. Some one brought a site to a friend of mine for help, and it needed some. For example, the title element for the home page used over 10,000 characters! But the amazing thing was that the page was ranking in the top ten for the last three words in that title, and they occurred nowhere else on the site or in backlink anchor text. This was not a highly competitive search, but it was not without decent traffic, either.
I wouldn't have believed it if I hadn't seen it for myself - it sure wouldn't have worked well on the old Infoseek, you know?
So I took that as a sign that everything is not as we assume. And since that caught my attention, I'm noticing many examples in the SERPs where those "traditional" factors such as keyword prominence and density in the title element don't seem to have the power or importance that they used to.
[edited by: tedster at 8:13 am (utc) on Jan. 21, 2008]
| 7:05 am on Jan 21, 2008 (gmt 0)|
wow guys.. as the original poster I'm really enjoying this
| 9:45 pm on Jan 21, 2008 (gmt 0)|
Good stuff Tedster. Thanks for the insight. I was wondering the same thing myself. I have used H1 at the top of every page and was thinking of changing to H3 tags and modifying it with my style sheet.
| 11:01 pm on Jan 21, 2008 (gmt 0)|
Why would you even consider that?
| 11:20 pm on Jan 21, 2008 (gmt 0)|
|Part of my thinking here also comes from noticing that source-ordered content (using CSS positioning) doesn't seem to give the same edge that it used to. |
I might have to disagree with that one tedster but I don't have any explicit testing. Everytime I've moved a site from tag soup to semantically structured source ordered content, there is a definite noticeable improvement shortly thereafter and from that point forward.
I've never tested taking tag soup and leaving it as is and then source ordering that. I would see no reason to do that other than to test the theory of SOC and its effect on the indexing of documents.
| 4:15 am on Jan 22, 2008 (gmt 0)|
I think the real test needs to be comparing content that is semantically well-structured but "natural order" with semantically well-structured content that is also source-ordered. No argument from me that tag-soup is a deterrent to ranking well.
Because of the track record that source-ordering has had for me, I would be very reluctant to ask a client to do less than that, and I don't currently have the bandwidth to launch any new test sites of my own. But in recent months, new pages that are source-ordered do seem to me to have more of a challenge than they used to.
It's just a gut feeling right now - I'm noticing there are more finely nuanced semantic elements in the relevance metrics now that go beyond things like sequence in the source code. Google seems more able to zero on content anywhere on the page. And that's a good thing, IMO. The accuracy of some of the long-tail traffic I'm seeing lately is amazing, and that takes some major IR chops to pull off.
| 4:13 pm on Jan 22, 2008 (gmt 0)|
I recently read an older post you made a while back on the topic of H1 text tags v. images within an H1 tag. I believe you said at the time that text trumps images (bad news for the graphic designer wannabes like me).
But now that Google is more sophisticated, do you still believe you lose SERP points if you jazz up your web pages with nice fonts instead of plain Jane text?
<h1><img src="coolfont.gif" alt="Boring Font"></h1>
| 4:42 pm on Jan 22, 2008 (gmt 0)|
In my less experienced days I used an h3 at the top of a page because it gave the look that I wanted by default. It didn't reach page 1 in the SERPS for major search arguements. Then I discovered css and used an h1 styled how I wanted and now most of those searches are in the top 3 and the rest still on page 1.
| 5:53 pm on Jan 22, 2008 (gmt 0)|
p/g, yes you're still operating at less than full power with an image and an alt attribute. If you want to use cool fonts for headings, go with a substitution scheme - sIFR (Scalable Inman Flash Replacement) works really well.
| 7:57 pm on Jan 22, 2008 (gmt 0)|
Okay, I need an education here. What exactly is semantically structured source ordered content?
| 9:18 pm on Jan 22, 2008 (gmt 0)|
Semantically structured means you are using html tagging exactly "by the book" - your headings are used as described above, the bits of your content are all in appropriate container elements, and html mark-up is not used just to get a certain appearance in the browser.
For example, if you use <blockquote> it actually holds a quotation and is not used just to get an indented paragraph. Another example: text is not just lying around inside the body tag but uncontained by appropriate mark-up that communicates its semantic quality, such as <p> for a paragraph. Tables are used for tabular data, not to create the layout. And so on.
Now on to "source ordered". We're talking about using css positioning to put the elements where they belong in the browser window. When this is done well, you can put the most important content first in the body tag, even though it will probably display on screen below the header section, main menu, etc.
[edited by: tedster at 9:18 pm (utc) on Jan. 22, 2008]
| 9:18 pm on Jan 22, 2008 (gmt 0)|
My example above is very generic but should give you an idea. Semantically Structured meaning to use those elements available to me that provide "meaning" to the content they encase.
Source Ordered Content meaning to place those elements in an ordered fashion using CSS and AP (Absolute Positioning). This allows you to move the core content of the page right up after the opening <body> element. From my perspective, it gives the page "meaning" immediately without any guesswork. It also makes for a nice user experience with styles turned off and since that is how the bot is browsing... ;)
P.S. My footers are typically within the main content so it is not absolutely positioned. It flows with the content.
[edited by: pageoneresults at 9:47 pm (utc) on Jan. 22, 2008]
| 9:23 pm on Jan 22, 2008 (gmt 0)|
Also, since the critical content is served first, that important text is usually visible in the browser extremely quickly -- even while other parts of the page are downloading. Very good for keeping many visitors happy, no matter how slow their connection and no matter how much trouble the rest of their packets may be having getting routed appropriately.
| 9:26 pm on Jan 22, 2008 (gmt 0)|
| 4:04 am on Jan 23, 2008 (gmt 0)|
I'm sure I'm not alone when I say that I have no clue what semantically well-structured but "natural ordered or source-ordered" is. Can you clue us in?
| 4:14 am on Jan 23, 2008 (gmt 0)|
I gave an answer a few posts above - what do I still need to clarify? Natural order would be writing your source code without using CSS position rules to position the content blocks on the page - that is, just letting the browser flow it naturally.
| 7:18 am on Jan 23, 2008 (gmt 0)|
I have done a before-and-after comparison on a site that was semantically structured and naturally ordered, but set up in tables with some nav links up at the top. We went to CSS positioning and moved the nav links down to the bottom. I noticed no change in rankings, but I'm not willing to generalize from this and say that CSS positioning doesn't make any difference. This was just about a year ago.
Some years back, on the other hand, prior to CSS positioning, when I had to use the "table trick" to get the left nav out of the way, I would often observe a one or two-place move up in rankings after the table trick got the nav text out of the way (and put our H1 up right after the body tag).
One experiment that would be interesting, but more than I'd want to subject a client's site to, would be to change positioning of page template elements on some pages, but not on others, and see how Google reacts.
| 8:32 am on Jan 23, 2008 (gmt 0)|
What would be the programming difficulty for Google to identify the navigation links and then omit them from its ranking for a webpage?
I still don't know why we have to mess around with various tricks for positioning. It's ridiculous.
I think, despite how simple it looks, Google has had problems programming it, and that's why, as Tedster pointed out, you see navigation anchor text in snippets, not to mention odd page ranking related to the navigation links.
Google gets very confused. I suspect it got so bad at Google separating the navigation links from the content they decided to target sites with their 950 penalty. All those spammy links were throwing off their algorithm.
Tedster, thanks for the idea on fonts. I'd not heard of it but will look into it.
Even so I still don't see why Google is unable to see that a graphic image enclosed with an H1 tag and an alt tag that matches the title tag is not cause for concern.
I understand it doesn't like spammy alt tags which have been abused for keyword stuffing, but it should make a distinction between tags that match the page title and tags that don't. It's natural to match the H1 to the title tag, isn't it? So why would it be unnatural to have an image with an alt tag enclosed in the H1 tag with the same text as the page title?
It's 2008 and we're stuck with the same number of text fonts as 1998?
| 2:50 pm on Jan 23, 2008 (gmt 0)|
Let's put aside the potential indexing benefits for a moment and think about SOC from a maintenance standpoint.
I have my editor open almost all day, every day. When I'm in code view, guess what? I have very little scrolling to do as my core content is right there after the <body> element and I can start editing away.
The other advantage I've noticed is that I can easily grab blocks of content and move them around to my desire, I have no limitations other than those imposed by my lack of CSS knowledge. ;)
Since working with SOC, snippet generation has been much cleaner and succinct. I don't end up with all the typical garbage that a tabled non SOC layout is going to be up against. I don't have 2000 lines of code sitting after the <body> element that is nothing but top and left navigation.
Unfortunately we've hijacked this topic to the point where we might want to change the title. ;)
| 6:21 pm on Jan 23, 2008 (gmt 0)|
Thanks, pageoneresults and tedster...
| 2:40 am on Feb 15, 2008 (gmt 0)|
*** I have very little scrolling to do as my core content is right there after the <body> element and I can start editing away. ***
Me too, to a certain extent, but only because the next thing after the <body> tag is sometime like:
<?php include("/header.inc") ; include("/navbar.inc") ?>