How important are heading elements to the rankings of webpages by search engines?
I’ve seen arguments by people who write about and study search engines and SEO very closely, which often appear written up in “SEO Expert Ranking Lists,” that HTML heading elements (<h1>, <h2>, etc.) are very important, arguments that heading elements were once important and are no longer, and arguments that heading elements were never important. Sadly, all of those arguments are likely wrong. Not so much about the importance or lack of, but rather about the reasons for that importance.
A search engine might notice when a word or term or phrase appears near the top of a page, or above a wall of text. It’s also possible that a search engine pays attention when those are shown in larger font sizes, or bolder than the rest of the page text, or a different font than the remainder of the words on the page. But that prominence and that display aren’t really what a heading element is about. HTML has a font size large attribute and property. There’s also a bold property. Any words on a page near the top of that page might be said to be more prominent than others.
You can use any HTML element attributes and values and/or cascading style sheet properties to make words within different HTML elements bolder and larger and to transform them to all capitals or a different font or color, or all of those if you want. You can purposefully place certain text at the top of a page to make it appear that the rest of the page is described by those words.
A search engine might see that only a few words are bold on a page, or are italicized, or in a different font or color or larger size and take some kind of meaning from that, perhaps even giving the use of that word or set of terms or phrase a little more weight on that page. And those words could be in a heading element, but they don’t have to be.
Semantic Relationships in Heading Elements
When you use a heating element, whether <h1>, or <h2>, or so on down the line, you aren’t just impacting the look and feel of the text within that element, but you are also defining a semantic relationship between those words and the words that follow them. You’re telling visitors, and search engines that the utterings on the page that follow are related to the terms in your heading in a meaningful way, even if you don’t quite understand that, and don’t quite do that right. And many people don’t.
When you use a top-level heading element or an <h1>, you’re setting up a semantic relationship between that heading and the remainder of the content on a page, describing what it is about. If you then use a second <h1> on the same page, you’re creating some potential confusion, because someone or a search engine might see that as the ending of the semantic relationship between the content after the first <h1> and the start of this new <h1>. If instead, you use a second-level heading element or an <h2>, you’re continuing the semantic relationship between the top-level heading above with that content, but defining an included semantic relationship with the content headed by the second-level heading.
Words within heading elements might help a page rank in search engines because they are displayed larger, or bolder, or in different colors than the text they head. I’ve seen the argument that a search engine might give weight to words contained in an HTML heading element because they might presume that the content on that page is being defined by that heading.
Weight of Heading Elements Defined by How Well They Describe a Semantic Relationship?
Heading elements can help a search engine understand the semantics of words on a page a little better. Search engines can go out on the Web and index pages and explore the relationships between terms within headings, and the content they describe within that index. They can look for similar relationships on all the documents within their body of web pages that use the same terms within headings, and see if there might tend to be some kind of co-occurrence of words and phrases and concepts within those matches of headings and content using those headings.
So for instance, you may have a page that uses a top-level heading element (<h1>) of “Cities in New York,” and the page contains information such as the names of several cities in New York State and information about those cities, and there may be a good number of other pages on the Web that use the same heading and contain many of the same city names and information and concepts. You may also have another page that uses the top-level heading (<h1>) of “Cities in New York” while providing information about New Jersey Cities.
The New York Cities heading element with the New Jersey Cities information might not carry as much weight with the search engines as New York heading elements on other pages that head content about New York Cities.
Are search engines using the semantic relationships between heading elements and the content they head as a ranking signal?
They might be, or they might not be, and that might depend upon how well heading elements tend to describe the content they head.
Other HTML Elements and Semantic Relationships
There are many other semantic relationships from certain types of HTML elements that search engines are looking at more closely, and have been for many years, and a number of those signals seem to be pretty useful. There are patents and papers and even actual services from the search engines that describe and take advantage of those semantic relationships. More on that in my next post.
We found that H1 tags carry almost no weight in some controlled experimentation we did a few years back: The Triviality of On Page HTML Tag Optimization. We used XML sitemaps and GWT to get a ton of nonsensical content indexed in Google while only varying the location of the searched keyword (also nonsensical) inside various tags. Only the Title tag proved valuable.
Bill,
I have often wondered about best practices for H1, H2 etc… I have read many different theories, looking forward to the next article and diving into the patent. So, do I understand correctly that you advocate only 1 H1 heading per page/post/article because additional H1’s will send an unclear signal to the SE’s?
Thanks for the reference Matt. Good Read
What an interesting post, Bill.
This is a real “back to the basics” write-up compared to what you usually do.
I have to say, you threw me for a loop with this statement: “[arguments that] heading elements (, , etc.) are very important, arguments that heading elements were once important and are no longer, and arguments that heading elements were never important. Sadly, all of those arguments are likely wrong”
But after reading your entire post, and please correct me if I’m wrong, you’re saying that heading elements increase in importance as the semantic relationship between them and the content they precede strengthens?
That would make sense to me. Actually, I could see this relationship being multi-level.
i.e. section content semantically relative to heading tags and page content semantically relative to the title tag…
It seems like this would increase the SERP values of relevant pages while negating the effects of keyword stuffing.
That makes sense to me and seems kind of obvious once you stand back and look at it in those terms…
Mark
Semantic relationships are easier to define now with HTML5 since it brings new ways to markup sections with elements like section, article, aside and nav. These create explicit sections. Without those tags h1-h6 would create explicit sections.
For further understanding of the document outline algorithm I recommend reading:
Sections and Outlines of an HTML5 Document
HTML5 And The Document Outlining Algorithm
HTML5 document outline revisited
Not sure how all this will affect SEO in the future but the tools to better mark up documents semantically are now available. That is, once you get past all the different arguments about what tag to use where and when.
I use the example comparing search engines to a college professor reading a thesis in the way a search engine might evaluate a Web document for relevance and possible ranking. As students, we’re taught to build an argument based on the preceding information, in a reverse hierarchical order, e.g. global to specific. Web pages are similar. The bot scans the Page title, H1-Hx tags, emphasized words (bullet and numbered lists, bold face, italics), semantic relationships of any number of key terms and external links (for supporting documentation).
A strong term paper would include all these elements as well. The professor scans the opening statement to determine if you have laid out your primary theme (keyword), subheads, (refined keyword concepts) bullet lists for more key concepts and summation (keyword wrap up.) Footnotes/resources provide proof of authority.
This is why writing Web content is a skill few master and why I write for readers first and then go back and optimized for search engines.
Heading elements may have decreased in ranking power over the years, but as a developer I still believe it is best practice to use them.
H1s and H2s can easily improve rankings in many queries — I test this regularly and have done so as recently as only a few months ago. They can also be as useless as 1,000 links with anchor text. The search engines pick and choose what they will show to users for reasons that simply cannot be reverse engineered by simple SEO tests.
I’ve been testing this at several occasions during the past 3-4 years and I have yet to see any impact at all of using H-tags in the text. That said: They do no harm and they might even have some significance in other search engines than Google.
Fabu post – as usual – Bill.
Historically, at least, there’s good reason to believe that headings packed a special punch (and, like Michael Martinez, I believe they continue to be helpful from an SEO perspective).
Which brings me to a favorite quote from that oldie but goody, Hilltop, which I offer without commentary:
To locate expert pages that match user queries we create an inverted index to map keywords to experts on which they occur. In doing so we only index text contained within “key phrases” of the expert. A key phrase is a piece of text that qualifies one or more URLs in the page. Every key phrase has a scope within the document text. URLs located within the scope of a phrase are said to be “qualified” by it. For example, the title, headings (e.g., text within a pair of tags) and anchor text within the expert page are considered key phrases. The title has a scope that qualifies all URLs in the document. A heading’s scope qualifies all URLs until the next heading of the same or greater importance. An anchor’s scope only extends over the URL it is associated with.
Source: http://ftp.cs.toronto.edu/pub/reports/csrg/405/hilltop.html
I can see one important use of the sub-headings (h2,h3…ETC). Dividing long articles into “chunks” of semantic information and using anchors within a “table of content” at the top for easy navigation pointing to theses chunks. In some cases these are called: “Jump links”. it seems only reasonable to use sub-headings compared to regular anchored text or emphasized bullets, even though it is not a must.
In the last year or two Google started to show “jump to ” snippet links inside the meta description. That’s one way to gain an advantage and make emphasis to a page in SERP’s for terms other than the few ones we had in mind in the first place (which may not be even included in it’s meta title for various reasons and space limitation)
Content drives structure. You develop the content and then add the structure for ease of comprehension. That is how I see it. So, if you just look at enhancing the content for search elements willy nilly with the smattering of heading tags, it will have the ‘right’ tags but miss out on the correct semantics.
One question: Are search engines really that smart?
Turns out all of this may not matter as Google is now saying the GOOGLE+ network will influence search results. What has been true for years may soon no longer be relevant. Links used to be gold will it be google+?
The amount of time it takes you to do research on whether H1s matter is better spent on actually writing H1s and paragraphs that make sense to your intended web audience
Google’s search results don’t work just one way any more. They haven’t for a long time. “Context” is not just the text surrounding text — it includes time, location, connections, word order of the query — all things that Google measures differently depending on where you search from and whether you are logged in and what your logged in/out settings may be.
Thanks Bill for bringing to debate this old but very interesting topic.
@Matt Mikulla: I will definitely read the links you shared to implement and pursue the benefits of new markup possibilities via html5.
Matt Mikulla (#3) noted it. There are new semantics with HTML5 (for W3C Fans) or simple “the new HTML” (See WHATWG Spec).
Semantics like or will define an outline and e.g. every article or section may have its OWN .
So heads up, start working on new HTML refactoring, since it will have great benefits for user experience AND will make you web dosument much more semantic!
Kind regards
Patrick
It’s amazing how many webmasters or so called “seo experts†still don’t know the proper use of heading tags. First, many people forget they aren’t just used to inject keywords onto your web page but rather to “structure†the content so a search engine and web visitor know what the web page is about. I understand that each of the tags (H1, H2, etc) is a different size and possibly appearance (bold or unbold, etc.). Does this interfere with the overall font and font size of the site? Good to see more blog posts aimed at the proper use of header tags.
The problem is, many SEOs are doing it the wrong way around, multiple times.
So many simply read up what others are doing, and emulate/replicate … with little/no thought as to what it is they are doing/why.
Others tend to look for “advantages” and “magic wands” … they want quick effects.
Very few actually stand back and look at what it is Google wants to do, nor how G may try to do it.
Once you start standing at the end of the race, looking back down the track … you tend to see a different picture of things.
Remember – G wants Quality Relevant Results.
There are only so many signals.
There are only so many data points.
So it boils down to how those things interact/accumulate etc. that is important.
If you have content that is non-semantic … G can/does handle it fine.
But, it will take it that bit longer to process/rank it.
I purchased an SEO WordPress plugin recently that rates the quality of your post and it seems to place a pretty decent amount of weight on whether or not the post contains , , and tags with your keyword in each of them. However, after reading this, I’m not a bit of a skeptic. At least they break up large chunks of text, but other than that, it seems as though they’re not as important as I once thought.
Interesting to think of a relationship between h1 and h2. Had never considered multiple h1s causing confusion. Would this be just for h1 tags?
Thanks for your article. I understand your point of view about multiple h1 but it is now a HTML 5 standard to have more than 1 in a page (smartly used of course).
Would you think it will be an issue if we use many h1 in a page for HTML 5 structure, including the first one is different on every page ? By the way, i think Google will use the entire context of your creation (title, subtitle, h1, keywords in phrases, incoming links, external links, incoming social signals, social signals origins…).
What do you think of this ? With all of this, i am not sure it will confuse a search engine having multiple h1 in the same page, right ?
I did a little experiment back in 2008 with H tags. For every page that I added these tags to that pages keyword phrases improved in the Google SERP’s within days. But this is now 2012 and I have not experimented since then. I would likely use them even if they had no SEO value for their ability to highlight what a pages content is all about.
If I am looking for Carpet Cleaning Service in Tucson and that the first thing that I read I am likely to stick around and look around.
Hi Russ,
With approaches like phrase-base indexing a possibility, I’m not sure that testing based upon nonsense words is going to be particularly insightful. I’d also be concerned with how Google might approach terms that are very rare on the Web, and the possibility is that they treat them very differently with more common terms and phrases. Under a phrase based indexing approach for instance, a very infrequently occurring term might even be relegated to a supplemental index where very little data about it is even collected.
Wow! This was FUN to read… It really sucked my ADHD Brain in, and that’s hard to do with an ADHD Brain! That being said,
I do claim to be an absolute SEO Genius within my own home town, but certainly not out here on the entire web, I’d get my but kicked! Besides, being an SEO genius isn’t too hard in a town with population 135! (ok, so I like prime numbers and sarcastic humor) 😉
But what I’ve found to be a very good fail-safe is to think as if I own Google and I want my customers (searchers) to be happy. So that means (If I was Google) that I must be able to find the page, I must be able to understand what the site/page is saying/about. Then, once I (Google) have figured that out, then I must decide if the site is easy to read and yummy to humans and rank it accordingly.
In a nut-shell, what you said in this article makes perfect, although impressively wordy, sense. Just make it easy to understand and easy on the eyes and easy to consume by both humans and robots. Also, if I owned Google and saw a page that was too perfectly optimized, I’d have to have a human to review it… So, sometimes I’ll make little SEO mistakes on purpose.
Hmmm, or would that also be too well thought out and optimized? (hehe, I crack me up!):P
Gary Anderson II
Hi Cody,
I would probably just recommend one h1 on a page, and treating it as the main topic of a page.
Often people use main headings more as a matter of formatting than of setting a semantic relationship between headings and the words that they are headings for. It’s hard to guess how Google might treat web pages in the wild that have multiple h1’s, because a strict interpretation would be to treat the different sections of a page that have a h1 heading as if they were separate semantic segments. On a Web where people might use multiple h1 elements more as a matter of getting the right look on a page, following that strict approach might not make as much sense. I’m more comfortable with Google getting it right if I only use one h1.
Of course, when we start talking about HTML 5, that changes significantly, since different semantic sections of a page can each have H1 headings. Hopefully when people start using HTML 5, they will have a sense of the implications of its use from a semantic perspective.
Hi Matt,
Thanks for the links.
HTML is so flexible, and there are so many different ways that you can present the same content in the same way with completely different code under an HTML page. HTML 5 does introduce more semantic markup, and a different way to think about headings. And I believe that people who work at Google have been involved in setting some of the standards for it, like Ian Hickson and Mark Pilgrim.
So I imagine that there probably have been a few discussions at the Googleplex about what HTML 5 might mean for search. I can’t recall having seen a patent filing yet that discusses the semantic implications of section, article, aside and nav elements when it comes to search. But that doesn’t mean that there isn’t someone working on one. 🙂
Hi Nicolette,
Great points. Where we sort of get away from that is when we have to start to account for headings, footers, sidebars, and other segments of a web page that make it a unique medium of its own. There are also many pages on the web that are formatted in completely different manners, and in some cases might be more like magazine covers or newspapers, or forum threads, and even social network pages that might have their own unique semantic structures as well.
For instance, it might be tempting to use a heading element in a sidebar to top off a list of links, or to introduce a sidebar widget or some content, and I usually recommend that people don’t use those headings there, if possible, for a number of reasons, such as those types of sidebar features often being repeated on other pages on the same site.
When we write a page that’s pretty straight forward, I think the analogy you make is a pretty good one. But when we start looking at some other types of layouts and think from a design perspective, it might help to consider the semantics of what we are doing carefully.
Hi Ryan,
The thing is, we don’t know whether heading elements have decreased in the way that they might have helped a page rank for a term or not.
I don’t recall seeing something in the phrase-based-indexing patents about headings, but I can envision a search engineer mulling over the idea that if a heading is a good one, then the content that it is a heading for should contain some phrases in it that might be related to one or more phrases in that heading. And if that happens, than that heading might carry more weight than one on a page where that doesn’t happen.
Regardless of that, I think headings also make pages more readable, even if your don’t explicitly use a heading element in your “heading.” They can effectively introduce a topic in a segment on a page, break the page up into reasonable parts, and make it easier to read. See, for instance, Jakob Nielsen’s use of headlines in: Writing Style for Print vs. Web.
Hi Michael,
I agree with you 100%. I do think there are many things that you can do from an SEO perspective that provide you with an opportunity to succeed. I often write here that I look at patents and whitepapers from the search engines not so much to learn about the methods and processes described in those patents as I do to learn about the assumptions and perspectives behind them.
From a semantic perspective, its even sometimes possible that the words in a heading on a page might not help that page rank as well for some or all of the words in the heading as much as they help the page rank for other terms on the same page that might be “related” in some manner to some or all of the words in that heading.
Hi Thomas,
I have seen pages that appear to benefit from the use of headings, and as I wrote in the comment above to Michael, it’s possible that the addition of a heading that doesn’t include the term or phrase being optimized for on that page could potentially help it rank higher for a query that includes that term. That’s a possibility under a phrase based indexing approach, for instance.
Hi Mark,
This post originally started out as an introduction to my post 10 Most Important SEO Patents: Part 7 – Sets, Semantic Closeness, Segmentation, and Webtables, but quickly developed a life of its own. That might be part of the “why” behind its back to the basics approach.
Yes, there’s a possibility that they might, but let me go at it again, from a slightly different perspective.
If you go through my Part 7 post, and the section on semantic closeness, one of the things that patent tells is is that words in a page title might be seen as being equally the same distance as every word on a page, and words in headings might be seen as equally distant as every word in a section that they are the headings for. So there is some kind of semantic relationship there that Google might be acknowledging, and the implications behind that are interesting.
If you read the section on Webtables from that post, you see that Google is looking at the headings of table columns to get an idea of how well those labels fit the items in the column under the heading for that column, and that they might look at lots of tables that might share similar subjects and attributes to come up with confidence scores for items with similar labels, and also may sometimes find semantically related labels for table headings that are missing in those tables.
It’s quite possible that Google could also be looking at headings on pages, and the content in the section being headed, and compare content in sections on other pages that are similarly labeled (with similar headings), and coming up with a confidence score about how well the heading fits that content. They could look to see if they share uncommon words, or phrases that might be related (under a phrase-based indexing approach), or both or other factors as well. If the heading has a high confidence level of being about the content headed, it might carry a higher weight than if there isn’t much of a semantic connection.
Hi Eitan,
Good points. I have to say that you had me racing to find this Google patent: Artificial Anchor for a Document.
The patent tells us that it might look on a page to see if there are already the kinds of jump links, or named anchors, on the page already. Jump links can work effectively as headings that break a page up into semantic parts. I was wondering if the patent had any mention of why it might choose some pages and some terms to treat as “artificial anchors” from a semantic perspective, but it didn’t. Still worth a look. 🙂
Hi Aaron,
Thank you.
And thanks very much for that reference to the hilltop algorithm, and how it might look at headings. I’m going to have to spend some time revisiting it.
Hi Kris,
I might take the opposite side of that (Content drives structure). I think the medium you use can have a significant impact upon how you present something. I often try to make pages scannable if I can, with headings, tables, images, and other things that make reading on a monitor easier. I find that coming up with headings for sections that I’m writing while I’m writing makes creating that content easier. I may change my headings during that process, but I find it makes it easier for me to define the ideas that I’m trying to get across by using headings and subheadings, sort of like the outlines we learned about in grade school.
Search engines aren’t necessarily that smart, but many of the people who come up with the algorithms behind search engines are pretty smart.
Hi Jacko,
Google’s “Search Plus Your World” might show you more personalized results when you want to look at them, by looking at people whom you are connected with in Google Plus have shared or endorsed, but it doesn’t necessarily change the search results that you see when outside of that personalization mode. At least not yet.
Those social signals might influence results that aren’t personalized at some point in the future, and things like Google’s authorship markup might influence that to a degree, helping Google to understand who may have created content on the Web, and who might have copied that content.
I think Google will continue to look at semantics more deeply in the future regardless of whether or not links might be worth less.
Hi Vince,
If I spend time researching the value of h1 elements in order to write about them in a way that makes sense to my audience, does that make things different? 🙂
Hi Michael,
We’re all just entities to Google, and the intent behind our searches can be very different if we’re surrounded by corn fields in Nebraska, or monuments in DC. Our preferred language, preferred country, previous query, web history, and the data center we are routed to, along with other settings, can influence the algorithms used for our queries and dramatically transform the results we see.
Hi Eliseo,
Not too sure how old the topic is.
I’d like to think my post covers at least a little new ground.
Hi Patrick,
I think HTML 5 does have the possibility of making people think more about semantics when they create pages, and I think that’s a good thing, too.
Hi Autocrat
Often when people learn SEO, they focus on the “how,” without trying to dig down deeper into the “why.” The only time “magic wands” ever work out is when you do all the work behind them that create the possibility of them having an impact.
Hi Laura,
Not quite sure that I follow what you wrote. I’m guessing that you might have tried to use a heading element in your comment, and it disappeared.
I don’t quite understand why you’re saying that you don’t think headings are as important as they once were, either.
Hi John,
Things might get confusing for a search engine that might be trying to understand semantics if you were to use heading elements out of order, too. A smaller heading element can’t contain a larger one.
But having two h2 elements in a row should be fine.
Think of the different heading levels like you might like the outlines you may have created in your grade school days, with main headlines, and indented subheadings. An outline might look like this:
Main headline
– Second level headline
– Second level headline
— third level headline
— third level headline
– Second level headline
— third level headline
— fourth level headline
—- fifth level headline
— fourth level headline
– Second level headline
I find that building content using H tags keeps me on the straight and narrow as I tend to veer off and try and include too many concepts otherwise. Following them down the page and making sure I’m staying on target. Like main chapters/sub chapters of a book. Found it interesting reading the comments, good links to interesting H tag info. Thanks!
Hi James,
Thanks. Ideally headings should be emphasized in some way to help them stand out as headings to people viewing them, and a browser will make them do that if a style sheet isn’t set up on a page to control how they look.
Hi Ramenos,
While HTML 5 is an emerging standard these days, most websites are still using older standards. The links that Matt pointed to in his comment are helpful when it comes to using h1 elements in HTML 5, and I would recommend spending some time looking at those if you have questions about how those are used there, and how the search engines might treat h1s in HTML 5 pages.
Hi Urban,
A page with a main heading that both stands out, and is completely on point with what you are trying to find is usually a good sign that a page might help fulfill your informational or situational need. 🙂
I’d still use main headings even if they had no SEO value as well.
Hi Gary,
Happy that you enjoyed the post. I promise that I tried to keep the post as short as I could but my fingers wouldn’t let me stop.
And make it make sense. Use the words on the page that you readers expect to see. Understand that when people read about certain subjects that there are likely some words that should appear upon the page that you’re writing that it makes sense to include on those pages. Search engines will try to identify those words by doing things such as looking at other pages on the web and seeing if certain words and phrases tend to co-occur, and being aware of that can help you.
Hi Charles,
Thanks. Using headings as you write content can definitely help you focus, and can lead you to ask more questions.
What are the concepts that this page should contain? What are the main points that I should try to convey? What things are essential to cover, and what things can I leave out? Is this topic that I want to include on this page something that should be given its own page? Again, it’s a little like the outlines that we were taught to use in school to help us write. Including headings like those help make our content easier for readers, and there may be a benefit from the search engines as well.
Thanks for the link Bill, very good read.
Thanks Bill for that extra morsel of guidance!
You said, “Use the words on the page that you readers expect to see……Search engines will try to identify those words by doing things such as looking at other pages on the web and seeing if certain words and phrases tend to co-occur, and being aware of that can help you.”
You’ve taken what I said.. and used it and expanded on it on a way that would connect with my though process(getting inside my head) to give me an even deeper understanding. And it worked! Thanks! That is not a bs compliment, I really mean it.
If you’re trying to get me to be a loyal reader, it’s working! haha 😉
Gary.
You’re welcome, Ryan.
I really enjoy reading Jakob Nielsen and Jared Spools articles, and while the two don’t always agree with each other when it comes to different usability issues, I always appreciate their perspectives.
LOL! I’ve never bothered at all with h1 elements when it came to improving my website’s search engine ranking, so it warms my heart that the general consensus about them here is that they aren’t much chop.
I think the tried and true methods of backlinking and posting original material are the things webmasters need to concentrate the bulk of their seo efforts on. Everything else is mere bunting.
Hi Obie,
I think h1 elements still have value, from a search ranking perspective, from a readability perspective, and as a way to make writing a little easier.
Hi Gary,
You’re welcome. I’m working on it. 🙂
Just be careful not to go too many levels deep in headings – if you do, don’t do them as header tags beyond H3. The study referenced here found that H4 and H5 actually hurt you slightly – lends credence to the “over-optimization” warnings the community used to talk about frequently – see table:
http://www.coconutheadphones.com/googles-secret-ranking-algorithm-exposed/
Hi Ted,
That research paper rubbed me the wrong way for a number of reasons, and I find myself not ready to rely upon the conclusions that they’ve reached.
Their inclusion of “keyword density” as a ranking factor bothered me because its definitely not a signal that search engines use. How could they discuss keyword density and not tf/idf? Was it because they don’t have access to term frequencies for all of the documents in the corpus of Google’s index, so they just ignored it?
They are also looking at the use of keywords in a meta description as a ranking factor in Google as well. Google doesn’t use meta descriptions as ranking signals. A well constructed meta description might help clickthrough weights, but Google will ignore poorly constructed meta descriptions a lot of times anyway, so even that tertiary consideration (which they didn’t raise) is moot.
I could go on, but I’m skeptical about how knowledgeable they are about search and search engines.
Having said that, I find that if I have to go into deeper levels beyond a h3 on a page, it’s time to start considering whether that content might best be served on another page. 🙂
Yeah, that study rubbed a number of people the wrong way, I ranted a bit about that in my posting’s comments recently.
What you’re touching on is the whole “correlation <> causality” thing, which is definitely a problem with these sorts of studies.
Keyword Density will be pretty closely correlated to TF/IDF though (they studied a small set of terms so it could definitely be skewed). I’m also a believer that Google uses CTR as a ranking factor so that could explain the correlation of keyword in Meta-Description to ranking. I see pages pop up immediately high for a week all the time as if Google is collecting relative CTR information, then settle lower – lots of similarities to “establishing a good initial quality score history” in the paid search world.
You could be right, just because it rains every time you wash your car doesn’t necessarily mean if the crops are dying you should just run out and wash your car. But in a world where the weatherman holds all the cards and isn’t telling, maybe it’s worth a shot 😉
Hi Ted,
There was just too much misinformed and sloppy thinking in that paper, and the irony is that while it’s titled “How to Improve Your Google Ranking: Myths and Reality,” it seems to be spreading more myths than reality, and doing so in an academic venue. It also oversimplifies how Google ranks results much too greatly.
One of the ways that Google Webmaster Tools can be used effectively to help improve click throughs is to identify highly ranking pages that get lots of impressions but very few clicks, and see if meta descriptions are actually being shown for the pages ranking for those terms, and if they are, rewriting the descriptions to something more engaging. Google might be considering clicks in rankings, though possibly tempering that when something like a Query breadth analysis might deem it appropriate.
Google publishes lots of whitepapers, lots of patents, and lots of blog posts about how their search may work. If those guys were serious, they would be much better informed before they published.
I have a question. On my main page my H1 happens to be an image. Any way to go about adding text? that’s hidden to the eyes but search engine readable?
As others have noted, h tags are there for the reader more than for the search engines, and the proper use of them should definitely make an article more easy to read. As far as the search engines go, we all know that we have no idea what Google or other engines will include in their algorithms in another two or three years. It’s a pretty safe bet though that, if there is any change in how they handle h tags, it will be to increase their importance rather than somehow penalize sites for using them properly. Therefore I still see no downside to using them and using them very well. At the very worst, our readers will be well served and will thank us by coming back. At the very best, some day the search engines will say thank you as well.
So far I have been following the same scenario for putting my targeted keywords in header tags from now I would closely analyze the relation shop between the content and next tag which I have used.
I agree with the comment made to Cody about having just one H1 tag on a page. I tried some experiments a few months ago and G punished me. I switched to using appropriate H2, H3. And seemed to complete my penance because I am back on page 1 of G’s organic listings. (adding video links in my site made a difference too).
Hey,
Even i agree with you. i do not think thatthe search engine has something to do with the hi or h2 tags. if the text is to be highlighted to get it notified by the google search engine it could be a different font style, size, or colour than the rest of the page. it is just the same logic we use by bolding the keywords in the article.
Very informative post. I was one of those who thought Title and header tags really made a difference. Great to get some clarification on the same !
I don’t know what to believe (I’m not a SEO myself) but I think breaking up your copy using sub-headings (especially if the copy is long) is essential for readers of your content.
I’ve only been doing SEO on my site few 6+ months but the thing I THINK I know for sure is that there is so much mis-information out there. I think google have a team of people who put red herrings all over the internet. Why would you believe anything they tell you. You ask a particular question on forums regarding SEO and you get ten different answers.
Google might give you a bit of information, but why would they give you any info that helps you possible cheat the system.
I am a one man band, doing my SEO. I only started to pay attention to this statement when I had a good site to build and it was. Build a good site, that people will want to come to, with decent content and get some good links coming in. Focusing making that site for people who are reading it, not the search engines that are crawling it. I know there are various different factors but as a one man band this is hard enough to focus on it’s own. So far it’s working, but it’s slow, painstaking and frustrating.
Hi RichardC,
You could possibly add alt text to the image, but the HTML standards anticipate the use of text within headings, and not images:
http://www.w3.org/TR/html4/struct/global.html#h-7.5.5
There’s absolutely no guarantee whatsoever that if you encase an image within a heading tag, that alt text for that image would be interpreted by Google as a heading.
Hi Lorne,
When heading were first introduced, one of the thoughts behind their use was that they could be used to construct an index of pages on the Web, and could be show off that way, but I agree that they still have tremendous value to the viewers of a page.
I see no downside to using them as well, and expect them to be around for a good number of years, even if they might be used somewhat differently in the next HTML standard.
Hi Alex,
Sounds like a good plan to move forward with. Including targeted keywords in headings is usually a good practice. Paying attention to possible relationships between those and the content being headed by them is definitely worth considering doing.
Hi Dr. White.
It’s hard to tell sometimes whether a change that you’ve made has had that kind of impact or if it were something else. But it’s good to hear that your pages did recover in rankings.
Hi Nelson,
The thing is, I didn’t write the patents that I point to in my next post that say that Google might be doing more than just looking at words in heading elements as if the headings only have value because they are distinguished from the rest of the text on a page because of font size or style or color. They point towards some kind of semantic relationship between a heading and the content it heads. The post is at:
https://www.seobythesea.com/2012/01/sets-semantic-closeness-segmentation-and-webtables/
Chances are that search engines do look at how text is displayed and if it’s emphasized in some manner, but it is possible that the search engines are looking at the semantic aspects of those elements as well.
Hi David,
Thank you. Hopefully I’ve given you some new reasons to think that the search engines might treat them differently. 🙂
Hi Brett,
If absolutely no other reason at all existed for using heading elements, that would still be a very good one.
Bill
WRT your comment re: using GWT to find low impression keywords and then work on their meta-descriptions – totally agree, very useful exercise.
If you’re really ambitious you could install Microsoft’s Adcenter desktop tool and use it to get Bing’s systemwide Clickthrough Rates for those keywords (too bad Google doesn’t publish these like Bing does) and then adjust using that as a baseline to see whether you’re doing worse or better than you would expect. If you don’t mind having it install SQL Server just to run it 😉
– Ted
Hi Darius,
It’s possible that Google might issue a patent as a red herring not only to potentially mislead people who might take that information and want to learn more about how Google does what they do, but also to mislead others who work on and build search engines. And it relies upon a very small percentage of the population who might read through patents to write about them, like me and David Harry and maybe a few more people.
It does cost a lot to patent something, it can take some time and effort to put one together, the audience isn’t very large, and there are cheaper and faster approaches that reach much larger audiences.
Focusing upon quality and building pages that your audience will want to see is a very good approach. The things that I wrote about here are good business practices that follow standards set forth by the W3C, which have both a display and a semantic component to them. Google has shown in a large number of patents and white papers that they try to understand the semantics of the Web, as described and defined by the standards of HTML.
In many ways, SEO is an attempt to understand the framework within which your website exists.
Hi Ted,
Thanks. I’ve seen a number of successes with that approach, and it’s definitely worth using.
Thank you for the suggestion about the Adcenter desktop tool. I’m going to have to look into it.
Hi Bill,
you are providing good information and useful for SEO expert.I think effective search engine optimization involves many factors. Ranking highest in importance is providing informative content with proper page layout and programming. Much consideration should be made to your proper use of heading tags. A good example of the h1 tag is the heading of this paragraph. It is used to define the content of the page, and for proper document layout. Whether is it a Web page or a newspaper, every written article should have a proper heading. On the Internet.
Hi Bill, HTML tags used near to the top of the page helps in ranking well, but what if the same HTML tag is used somewhere near to the bottom of the page. Does it helps in ranking well, if not, then is it harming the webpage SEO?
What if instead of HTML tags, I prefer to use CSS styles to make the Heading appear bold and of same size as compared to HTML Heading tags say ?
I have heard that patent red hearing thing before, and I just don’t buy it. That is a lot of work and money (as Bill pointed out) for limited return, as eventually people will figure out that the red hearing is what it is.
I have always found, build a real site, with real content, for real people…and Google will like you and move up your site. Build a site specifically to be ranked (Exactly not what Google wants), and you will eventually get the slap down from Google.
Just my 2 cents.
Hi Bill, i have a question on these tags. I generally write about Tech news and relative stuff. There won’t be many subtitles as you know. So i’ve started making the important points of the article as H1 2 and 3.
Also i bold and italicize the keyword in the post. so does all this relate to too much SEO that hurts my Blog SEO ?
I am still not very comfortable in using tags as i am quite confused where they would fit well !
Hi Bill,
It does suit the content sometimes but sometimes it does not go well. Will there be any issue if the sizes of h2 and h3 tags are same as the size of the paragraph.
If there is no issue in having the same font size for them then maybe the problem would be solved. But i am not sure if On page SEO will be affected by this or not. What do you suggest in making the sizes the same ?
Yes that true Bill. SEO is a very complicated issue for starters. I guess experimenting is the only way to win over it.
Headings are a vital part of OnPage SEO, I must say..
Hi Sanjay,
Thank you. I think it’s useful to use headings, and can be helpful to readers and to search engines, but I’m not going to say that every page should use headings. I think that’s up to the person who creates content as to their preferences for how they write, and what they want to display on their pages.
Hi Travis,
I agree. Focusing too much upon inserting keywords into the content your pages, even if it makes what you’ve created painful for people to read, and you’re doing something wrong.
Hi Amit,
Not quite sure that I can answer your first question in a meaningful way. Chances are that Google may be considering the semantic aspect of how headings are supposed to work, and if you aren’t considering that yourself, they might do something different than what you intended. A heading heads content, and if you aren’t using a heading to head content related to that heading then how is a search engine supposed to interpret what you are doing?
If you use CSS instead of semantic HTML elements like headings, chances are that Google will look at the CSS headings that you created, and try to treat them as if they were headings anyway if you use them like they are. The emphasis is on the word “try.” You might have a little less certainty that they will if you decide to go that route, but it’s possible that they will.
Hi Sameer,
Use headings on a web page as if you might use headings on a paper document. Does it make sense to use a main heading at the top of the page, and sub-headings for different sections of that page? Does the main heading adequately describe the content below it? Do the subheadings adequately describe the content that they are sub-headings for? If so, then you are doing things right.
Hi Sameer,
I can’t tell you exactly what search engines do, or how they interpret things. I can only tell you what I’ve read from places like patent filings (from those search engines) and from my experience. If you present headings in a way that might try to gain the benefit the fact that they are headings but simultaneously attempt to obscure that those are headings, it might be that Google will ignore them as headings. But I can’t tell you that with any certainty.
Hi Sameer,
There are a lot of basics of SEO that aren’t too hard, and they are definitely good to start with.
Testing and experimentation are always worth trying, and I definitely recommend that anyone learning SEO have one or more hobby sites that aren’t business critical that they can try things with, and experiment on.
Beard made a good analogy that I think I will keep the “college professor theory” in the back of my mind when doing future SEO.
Hi Jonathan,
I think it’s a useful analogy as well. We often get pretty close to the content that we create, and if we can step back and look at it from a differnt perspective, such as someone like a college professor who might rate and grade and analyze the quality and substance of students’ work, it’s probably a good exercise to engage in.
SEOMoz ranks H1 tags as low importance, just something to throw out there.. and as well meta keywords. Title and meta description are most important, as well as InURL keywords. They also stress the importance of using your long tail keyword 4 times throughout a document, that’s the magic number. Thought this would help.
-Ryan
Hi Ryan,
In the SEOmoz correlation studies, the focus of the study was upon whether or not the presence of text that was larger and bolder and possibly a different font style and type impacted the rankings of pages.
It ignored possible semantic impacts of the use of headers. For instance, if a page was optimized for a specific term, and that term was in a section headed by a heading with a specific term that was semantically related to that term, the page might rank higher than it would if there was no heading, or if there was a different heading. That’s something that the SEOMoz “rankings” doesn’t account for in any way at all. And yet, the importance of a heading isn’t that it’s bigger or bolder or a different font, but rather that it builds a relationship between itself and the content that it heads.
I had the chance to participate in the first of the SEOMoz Search Engine Ranking Factors, and it was an opinion survey of experts rather than a scientific study. It didn’t ask us to provide what we thought were ranking signals and how they worked, but rather to assign number of signals that SEOMox came up with. I honestly didn’t agree with a lot of the results. The followup ranking factor surveys were even more constrained in terms of how much participants were able to assign rankings and values to certain factors.
I’m not sure that those even say anything about meta descriptions having any ranking value. Google has stated repeatedly that they don’t use terms in meta descriptions as a relevance signal. Those may influence click throughs when seen in search results, and that has value. But they are definitely not “the most important:” ranking signals. It’s not very likely that inurl keywords have that much value either – I’ve probably spend a few hundred hours looking for anything from Google, whether patent or whitepaper or blog post or inteview or anything to find something that says that. There is a value in links using URLs pointing back to pages that use keywords, but that’s hypertext relevance in action. There’s a Google whitepaper that tells us that Google might use keywords in URLs whey the need to do a quick and dirty classification of pages, but not necessarily for purposes of ranking.
As for a magic number of how many times to use a long tail keyword on a page, there is no such thing as a magic number. Not sure that the SEOMoz document says that either, but I doubt you would find anything from Google that says that. It might be advice from someone participating in the SEOMoz survey, but it’s an opinion, and that’s all.
Thanks for the info, Bill. Always appreciate the testing that goes into this for the SEO community.