Showing posts with label Friday. Show all posts
Showing posts with label Friday. Show all posts

Tuesday, 19 October 2010

Image SEO Basics - Whiteboard Friday

A few weeks ago Danny showed us some of the basics for video SEO, a medium that may not initially seem valuable for SEO purposes. Well, Danny dispelled that illusion swiftly, with a little help from his friend Doc Brown. This week, Danny's out there alone but still manages to show us that words aren't all they're cracked up to be; videos can yield some great SEO value, too. Besides giving us proven and actionable suggestions, Danny also postulates on some experimental and potential ways to optimize images that may prove useful now and in the future.

Hello, everybody. My name is Danny Dover. I work here at SEOmoz as the lead SEO. On today's Whiteboard Friday, I'm going to tell you about the basics of image SEO. We found, when we were doing correlation analysis, that images and specifically the alt text that's inside of them are a remarkably well-correlated metric for SEO. Besides just being useful for people, images are also, it turns out, useful for search engines. I think part of the reason behind that is that pages that are well developed tend to also have images on them because it helps portray information in a way that textual based content can't do.

Let me go over some of the important factors with image SEO. Number one, I already mentioned this a little bit, is alt text. Alt text is the text that you provide for an image in case it can't be displayed. Maybe the image is gone or maybe someone is using a program that can't display images. This is the text that takes it place. So it makes a lot of sense from an SEO perspective that this metric is going to be important because it's the information you tell the search engines and other technologies what the image represents. With these, I recommend keeping them below about 140 characters. It's a rough rule of thumb. Also, have them be descriptive and in line with what you're trying to target for that page.

Number two is the file name. This works off the exact same principles. The file name is also information you give directly to the search engines and to other technologies to identify what the information is about. I would gander, if you will, that the file name is probably a rougher signal than the alt text. Alt text, from my experience, when it's there, which is not all the time, in fact, alt text is not included many times which is bad for SEO. But when it is included, it tends to be a clearer signal than a file name which a lot of times is just algorithmically generated by the timestamp, so it's just a bunch of numbers.

Number three is the surrounding text. I think a lot of people don't think about this when they're thinking about image SEO. The text around an image tells a lot about the image itself. This makes sense, right? You'll see a lot of times where images will be on a blog post and you'll have a caption describing the image. This is just another signal telling the search engine and other people and technologies what it is this image is about. The surrounding text, and that can either be a caption, like you've seen traditionally, or it can just be the paragraphs around the image. A lot of times an image will be used to supplement what the textual information is talking about. So the surrounding text is very important.

Fourth, as with all SEO, inbound links are important. It wouldn't necessarily be inbound links to the image URL, although it could be, but what I mean in this context is links going to the page that has the image embedded on it. Just like in normal SEO, the anchor text of those inbound links and where they're coming from and how many of them are all really important factors for image SEO and then SEO in general.

Last is number five which is human categorization. The search engines, especially at the beginning when they were developing this image recognition software, used humans. They would hire people and they'd say, "Label this." Google was semi-famous for creating this game, Google Image Labeler, which I think you can still find online, where it would show you an image of, say, an apple. They would ask you in Family Feud style, which is a game show here in the States, to list words that are associated with that object. You'd say something like apple, and you'd earn points if someone else also said apple. Maybe it's red, Fuji, or Grandma Smith, or whatever it is. So other words that are associated with the image. And that way they could train their software to start to understand what general shapes and ideas mean within images.

On the other side here, I have some more theoretical things that search engines may be using, while the things on this side are the things that we know they're using. We've heard search engineers talk about this. We've seen direct evidence. These are things that I think you should pay attention to but probably just going forward. It's more just for your knowledge rather than for you to use on your day to day.

The first one is OCR. OCR stands for optical character recognition. It's a very established software. It comes in a lot of Adobe products. You can get it in lots of places. What it does is it scans an image and can identify characters in it, characters like letters or numbers or spaces or whatever. From that, you can take actual text out of images. Again, this is a very popular software. It seems very likely to me that search engines are using this at least to some degree. It would be very costly for them from a resource perspective to use on every image on the Internet, but it would certainly make sense if they were using it on some or at least playing around with the technology.

Number two is color analysis. It's very easy from a development perspective to identify at least one color, maybe the primary color, within an image. You pick a pixel and you see what the hex code or whatever it is that you're measuring that on, it will be based on file type. It's pretty easy to get a general idea of what the color of an image is. This is helpful from a design standpoint if you're looking for certain color themes that go with each other or color patterns. Now we've seen this actually in the SERPs, so if you go to Google image search, you can see now, and Bing actually had this first, you can go to the image SERPs and you can actually pick to see only images that are of a certain color. Black and white is the obvious one, but then other colors as well.

Number three is file size and type. This one, I think, is more all about the extreme. If the image is ridiculously big, it's probably not going to get indexed just because the search engines don't want to spend the resources on that. The exception to that would be if it's ridiculously well linked to also. It's about finding these outliers. You probably don't want to have an image that's really, really big. It's probably not going to get indexed. Again, I think what it really comes down to is this is hurtful for users also because they're going to have to spend time downloading that. If bandwidth is a concern, they're probably going to click away to begin with. Image size and along with image type, the standard image things are all probably fine for Google.

I've heard just a rough rumor here that JPEG is preferred, but honestly GIFs and PNGs and all those other things are probably fine. I would not worry about those aspects. Only worry about it if you're using obscure file formats, which you shouldn't be doing to begin with.

The last one on here is the other images on the page. This is twofold. The first part being the other images on the page are likely related to the given image and that's because they're on the same page. Right? The other part, and I see this happen a lot especially with bigger clients, is when you put lots and lots of images on one page, like an image gallery, those pages tend to be very hard to get indexed. The reason for that is there's not a lot unique textual content. A lot of times it's just overwhelming to users. It doesn't provide a lot of benefit in a search result.

That's all the time I got today. I appreciate you listening to this. Please feel free to ask questions in the comments below. Thank you.

If you have any tips or advice that you've learned along the way we'd love to hear about it in the comments below. Post your comment and be heard!


View the original article here

Monday, 18 October 2010

Making "Reference" Content Rank Well in the SERPs - Whiteboard Friday

I've always liked encyclopedias; when I was in middle school I started using Encarta on CD-ROM, and sure, I usually needed it for "help" with my homework, but sometimes I would stray to non-copy-and-pasting-from-encyclopedia activities and watch terribly animated videos of war battles or Shakespearean plays. My poor children will never know the joys of a succinct five page article on the American Revolution with an accompanying 30-second 160 X 200 resolution video! I suppose they'll have to make due with the way too informative Wikipedia article and an accompanying overly high-def retelling of events - do they really need to be able to see Benjamin Franklin's hickeys?

Anyways, if my aforementioned future kids do end up needing to write about the American Revolution, and you have a great site about it, how can you make sure they end up seeing your content? There are a lot of reasons for why it can be hard to rank for reference content, but fortunately, Whiteboard Friday is here to help! This week, Rand discusses some great ways to get your reference content to the top of the SERPs.

Howdy, SEOmoz fans! Welcome to another edition of Whiteboard Friday. This week we're talking specifically about reference content, which is a type of content that often has a tough time earning external links, has a tough time getting rankings and visibility in the search results. Yet a lot of people are both (a) interested in it form a searcher perspective and (b) have marketers who are interested in ranking for that type of topic so that they can draw in traffic to help brand their site to sell advertising, to build themselves up as industry authorities, and sometimes even to make direct sales as they relate to that reference content.

So, let's start with some tips, some specific action items that you can take that will help your reference content get more rankings. When I talk about reference content, I mean everything from, like, dictionary-type definitions to encyclopedic types of content to how-to content. Anything that is sort of less about a news item, an exciting development, or a blog post and more like a piece of content that is simply informational in nature and designed to provide sort of an evergreen long-term resource. It's tough to get this stuff ranking, but I think we can help.

First off, let's talk about keyword usage. As you're building out this content, a lot of people think, "All right. I need to have a certain number of the target keywords and I'm going to use these keyword variations and I'm going to have this keyword density." I talk about the keyword density myth a lot of the time. The problem is, and I think one of the reasons it doesn't resonate with folks or why people still say, "You know what, I think Rand is full of it on keyword density. It totally works," is because it is true that in many cases you can have a keyword that is used a certain number of times, a small number of times on a page and you can increase the number of times that it is used on a page and see the rankings go up. People say, "Well, that's proof that keyword density works." In some semantic form, that is technically correct.

The problem is density itself is not necessarily, is almost certainly not the metric. Let's say very certainly not the metric that search engines are using. So when you use that metric you might be conflating different variables. It could indeed be the case that adding more keywords and increasing technically what could be measured through density is helpful. But density itself is a bad way to measure things. What I'd urge you to instead think about is, "Am I hitting all of these items, and am I doing a good job with them?" If I am, chances are good that increasing my density, measuring my density, is going to add no value. Certainly, it's the case that the search engines don't measure it. We don't want to be doing things that are sort of obviously known to be not used by the engines.

So, things like using the keyword element in the title, preferably at the beginning of the title, particularly for reference content is really good. People want to see right in the title in the search engine results that your page is about the content that they're searching for in the H1 headline. The H1 headline may not help all that much, or specifically using the H1 tag to designate your headline, as opposed to just having it big, bold, and at the top of the page, may not help that much. So if it is a pain, I wouldn't worry about it. But if you can, it is sort of a nice, good semantics thing to do. Good web standards.

Certainly, having it in the headline, whether you're using the H1 tag or not, is important because when someone clicks on that result and reaches your page, you want to reinforce the notion right there at the top of the page, in the headline, that this content that they've reached is about what they searched for and it is what they just clicked on. When you have the disconnect between those words and phrases, I really worry that a lot of times your bounce rate will increase, you'll see people leaving the page. It makes good sense form a usability perspective.

The meta description is certainly a good place to use it. It will get bolded and highlighted in the search result. Even though it doesn't directly help with rankings.

The URL, same story. Although URLs do seem to have some nice correlation. It looked like in our ranking models that they have some causation influencing that. Certainly, you can see that when you change over to search friendly URLs that use the keywords in there those are very nice for SEO purposes as well.

The body tag usage. This is where people get super obsessed with keyword density. Most of the time, unless your article is really huge, I don't worry very much about keyword density or the number of times you use it. You use it a few times, you use it the number of times that it makes sense in the document -- two, three, four, five, six, right. Those are fine. But I wouldn't obsess about like, "Okay. Wait. I think we have it nine times here. We should only have it eight because the average of the top ten is that they're only using it X many times." Get out of town. Like, no way, man. This stuff is not helping here. It is good to use it in the body tag.

It also is surprisingly good to use it in things like the image Alt tag and in the file name of an image that's on the page. I don't know what it is. It could just be correlation. It might not be causation, but it turns out that the image Alt tag is higher correlated than H1s are. So, maybe it's just the case that people like having images that are on the topic. Or maybe the search engines actually do have a preference about this kind of stuff.

You should definitely be worried about readability. If a normal, average user comes to the page and they read it, but the material is not connecting with them and doesn't make great sense, get out of there. It's trouble. This is one of the ways that SEOs and people in independent websites can really compete with Wikipedia, which is oftentimes hard to read, hard to parse, hard to understand, not tremendously well written. It's written by a group of authors a lot of the time. A lot of the material can be dense. The same goes for a lot of professionally published content that just isn't as accessible.

Completeness. So, one of the things that I definitely think about and this relates back to sort of topic modeling and LDA stuff to whatever extent that's being used. Certainly it seems like it is being used to some substantive effect, but we don't know exactly how much. Being able to comprehensively cover the topic that you're talking about will mean that more people like your content, reference it, use it, enjoy it, share it with their friends, and it means that they are getting value out of it, which means that metrics like time on site and browse rate will go up, which might help your SEO, might not help your SEO, but will certainly help your site metrics. You care about those, too.

Then, I think a lot about the angle that you're taking with your writing. Things like, I'm going to take a research-driven angle, or I'm going to take an opinion-driven angle, or I'm going to take sort of a showing all the different controversial sides of this, or I am going to walk through the history of this. Having that angle that is sort of unique and people say, "Wow, when I visit SEOmoz, I feel like I get a really thorough understanding of all the issues around a particular topic. Or I get a very opinionated piece from Rand about what he thinks about a particular SEO tactic and how people have used it. Then I get different sorts of opinions in the comments." That angle that you take can brand your site, brand your domain, and your company as having useful information on that topic. All of these things are far better to think about. If you nail those, you're going to win out over keyword density.

Next item that you do have to worry about with reference content is architecture -- internal architecture and internal linking. We talk about this ideal link architecture, the ideal pyramid, a lot. You start with your home page. If you can do this thing where you've got a hundred links approximately-ish per page, a hundred unique links, and that's linking down to the second level with all of your categories and each of those are linking down to subcategories, you can get to a million pages in just one, two, three hops. Three hops from any single page on a site to a million subpages means that even the most robust quantity of reference content can be reached in a small number of clicks. That portends really good things for search engines and for users who are trying to parse through your material and potentially surf your site.

This is a great way to think about organizing your site. You're never going to get to this perfect layer, but if you can think about this organization as a structure as you're planning, it would be very helpful. You don't have to do this with your home page either. If you think about something like a sitemap, an HTML sitemap on your site that you link to in the footer of every page and that page links to all of these and then they all link to these, you've accomplished the same thing. You've basically made it three or maybe four hops from any page on your site to a million pages. That's a really good thing.

You should also be thinking about things like using categories and subcategories intelligently. You can't just be listing content. Those categories and subcategory pages have to be useful and valuable in and of themselves. We've talked about that a little bit in the past here on SEOmoz, too. The relevance and usefulness of those pages is going to predict whether they themselves can draw in links. If these pages can draw in external links, you know that's going to help all the pages that they point to down below to rank better, to earn more linkages and page rank and trust. Those metrics that will flow down through a site.

I think it is very important and very wise to look at models like what Wikipedia has done and NY Times has done, what About.com has done, with cross-referencing content at deep levels. When you get to these deep pages down here and it has a link back up to that category and over to this page which it's referencing in the content, that's super useful from a visitor's standpoint because they'll click more. You might have a higher browse rate, a higher pages per session, as well as driving SEO value in that the search engines might see this one or see that this is linked too and then follow those links out from there, pass more link juice and more crawling power across those pages.

The last one, and I know the most challenging one, is earning external links. Reference content, are you kidding me? It just doesn't get linked to, you know. How are you going to win with this stuff? But there are ways. Successful companies have done really good things on this front. The first one I recommend is from the content perspective. Multimedia content, visual explanations, these kinds of things rock. I was pointing today on Twitter to a post from King Arthur Flour. Can you think of a more boring company? King Arthur Flour? Are you kidding me? They have an amazing blog. Their blog has earned hundreds, thousands of links because they've produced these blog posts that are sort of reference content about how to bake French bread and how to do no-knead bread. What they do is make them highly multimedia intensive. So, every step of the way they've got photo after photo after photo after photo. Tons of comments. People just loving it to death. Granted, you know, they're in a moderately interesting area of recipes, but it is super competitive, and yet they rank for this stuff. They're able to draw people in. And they can show off the fact that, you know, King Arthur Flour is sort of very highly rated for this kind of thing by other professional chefs, etc. Those visual explanations, the video content, they rock, right. You're watching Whiteboard Friday, huh?

Next piece that I really like is doing things with research content as well as like charts, graphs, and data. Even if you take your data from third party sources and you reference back to it, if you're the one who produces the actual visual chart, other people who want to embed that chart, want to use it in presentations, want to use it in blog posts, who want to talk about it, are going to use your materials. You can check out the SEOmoz free charts section where we take a bunch of data that's from sources like Eightfold Logic and comScore and Nielsen and Hitwise, put them all together, and then put them into interesting charts that other people can reference and embed on their pages. Of course, they'll link back to those original sources, as well as to us. Those are great ways to get your reference content to actually earn those links.

The last one, two methods to kind of go out there and do distribution. Those are licensing and translation. These tactics are ideal because you'll see all these other sites that are copying your content are linking in to your work, referencing back to that original. That is going to provide for the fact that even though these might be technically duplicate content, when the engines see them referencing your single source, especially multiples referencing your single source, they're going to know this is the original. You can do this with licensing where you say, "Hey, I know you are in this industry and you'd like to license out some content. I'll be the reference resource for you. You can put this stuff on your site."

It's brilliant, too, for translation. As the Web is getting more global, more people are interested in this. More people are trying to rank for search content in all sorts of other countries. You can say, "Oh, buon giorno! Would you like to translate this piece into Italiano?" Right? Those kinds of things are absolutely phenomenal.

By the way, I had a great time in Milan with some friends from WebRanking.it and Marco, exceptional experience. The Social Media Conference there had 25,000 people come to it. It's insane. People care about SEO overseas, and you can leverage that to get these translated articles out there on the Web and then to have the links point back to you. What does it look like to Google when ten sites from all over the world are all pointing back to your reference articles? It looks like you're going to win at SEO.

All right, everyone. I hope you've enjoyed this edition of Whiteboard Friday. I hope you'll join us again next week for another one. Take care.

If you have any tips or advice that you've learned along the way, or if you also love pretty much anything HBO produces, we'd love to hear about it in the comments below. Post your comment and be heard!


View the original article here