Google I/o 2009 - Site review by the experts

Cutts: So if you have a site that you want to throw up here-- we got one on a piece of paper just before the session-- if you have a site, you know, write it down, bring it on up. We'll be looking at it throughout the course of this session. If we don't get to look at your site, feel free to catch me or Brian or Greg afterwards. So it's a good chance to introduce, folks, Greg Grothaus, Brian White. We're all members of Google's search-quality team. So our job is to make sure that if you type in your name, you don't get *** back. Right? Unless you're a *** star. If you're a *** star, then you want to get *** back. But if you just type in, you know, "iPod car" or something like that, you want to get a really relevant result. So without further ado, we're going to start jumping right in. If you have a site you'd like to get reviewed, again, just write it down, bring it up. Brian or Greg will start looking at it. I'll start looking at it. Yeah, bring up a business card. Anything like that. Okay, cool. Excellent. All right. And worst case, we'll catch you afterwards. Okay, so this one was kind of interesting. iCarKits.com. This was a submission in advance. Is the owner of iCarKits in the room? Excellent. So I can be mean if I want to. I can be sarcastic if I want to. There was some good stuff, and there was some bad stuff about this site. Mostly good, though. When you look at it, what do you think this site is about? What's the impression that you get? They put iPod kits in cars, right? That's pretty clear. So it's a nice thing that you know what does this site do. We'll look at a site or two later on where it's like, why does this site exist? We're not quite sure. And in fact, normally I do this at SEO--search engine optimization--conferences. And I get a lot of really bad ones. Like cheap-hotels-biz-discounts- online-4you/*** or something like that. And all the submissions were pretty much white hat, so I might throw in a few spam results that we've seen, just to kind of spice it up a little bit. Now, one thing you might not notice: I'll bet whoever did iCarKits.com is proud of the fact that-- do you see this little iPod? That's actually active. It's searchable, which is kind of cute, but you might not realize it, right? So let's pretend I have an Audi. And let's pretend that I have, you know, an Audi A4. So you can click through, you can pick a random model year, and then for that, do I have navigation? Sure, yeah, why not. I've got navigation. For that, you can see the products that you have. That's pretty cool. I haven't actually seen that sort of active scroller within a picture, right? That's kind of engaging. But do you think an average user, who's not necessarily savvy, who lands on that page knows to go looking at the iPod? Probably not. So one of the things that we always try to communicate is provide multiple ways to get to a page. If you can have drop-down boxes, these sort of clickable search stuff, search engines might be able to crawl through that. But it's really nice if you can have static HTML links, because the crawler pretty much only knows how to crawl through static HTML links. Now, Google actually can read JavaScript. And not just stupid, look for a URL in the JavaScript. We can sometimes click and execute and try to do really neat things with JavaScript. But not every search engine can do that. You don't want to make it hard for the search engines to get to your site, so you don't want the search engine to have to fill out a form to find these products. But once you land here, it's actually pretty good. Now, one thing that I noticed-- you know, they've got text phone numbers, so that's kind of nice-- a lot of people will put their information in an image, right? "Free shipping." See this "free shipping"? That's an image. Unless you said "free shipping" in the ALT tag, there's no way we can actually read the content of that image. Now, there are people at Google who have said, "We should OCR the entire Web." Right? We've talked about it, right? Google Books. We do OCR our own stuff. It would be really cool if we should OCR "free shipping," you know, and "enter the discount code," and you could index that. And then if somebody searched for "iPod car kit," with "free shipping," we might return that. But don't rely on search engines to OCR the entire Web, 'cause it doesn't happen right now. It's not going to happen anytime soon. So the more you can have nice links, like here's the phone number just in straight text, the better off that'll work. What else do you notice about this? Anybody notice anything strange about the URL? The URL is "advanced_search_result.php? "set_questions=1 sort=4d &search_question[7]=14." That's a pretty hard URL to crawl. If I were designing this site from scratch, I would probably make it iCarKits.com/products or catalog/audi/a4/2006, right? Because you can encode all those parameters in a very nice way that's sort of like a tree structure, and anyone could click around. As long as you have these sort of, you know, multiple parameters on the URL where it's almost like search results, it's a lot harder for search engines to get through and try to find those sort of sites. Does that make sense? So if you can make something like a very static, nicely crawlable link or set of links, that's going to be a lot more useful. I don't want to dwell on this one too long, but in general, it's a pretty good site. It's crawlable, you can click around on links, and you can usually find stuff. And Google does a pretty good job on it. But I did want to mention one little tidbit. You're developers, right? How many people here are responsible for a Website or a blog in some way? A forum? Okay. Why do you run your Website? Say again? For people to look at it. Fair enough. But why do you want them to look at it? What do you want them--what do you want them to get out of it? To make money. Exactly. This guy doesn't want to educate the world about how to, you know, install iPod car kits. He wants to make money. And that's okay. He can provide a great service and great information. So ask yourself, what do you want out of your Website? I happen to run a blog. I don't run any ads. I often ask myself, "Why do I run my Website?" Like, what's the point? I just throw stuff up there on the Web. I don't get anybody to subscribe to newsletters or anything like that. So you should ask yourself, what are you trying to get out of your Website? This guy is trying to get sales. If you look at the title of the Website-- I'm gonna hover over it in Chrome-- iPod car adapters-- 2006 Audi A6-- This is pretty good. You notice he has the word "iPod car." I would bet you dollars to doughnuts that probably his top phrase is "iPod car." A lot of people who are caring about phrases, they'll put that right there in their title. And I would probably bet you that he is getting maybe hundreds of visitors a day, which is pretty good. If you can get a few of those to convert at, you know, 180 bucks a pop, that's not bad. But one thing I noticed when I looked at this site last night is he doesn't even mention the most popular product for the iPod and cars. So let's see if I can bring it up. There's a Google AdWords tool, and I'll warn you in advance that the Wi-Fi in this room is a little bit spotty, so I tried to preload as many tabs as I could. But if you go to this tool, which you can just search for with "Google AdWords tool" or "Google keyword tool," what it will let you do is it will let you type in a phrase like "iPod car." And whenever I did that last night, it said, "You know what? People type in 'iPod car,' like, 100 times a day." What do you think they type in 250,000 times a day? There's a product called iTrip. It's from Griffin. And what it does is it turns your iPhone into a little FM transmitter. You know, it transmits over the radio. The word-- It's starting to come up. The word "iTrip" does not appear anywhere on this site. So if you're a developer, if you care about making money, you know, do a little bit of research. What are the words that people are going to use to find your site? So I solve my little CAPTCHA. I type in "iPod car." And eventually what it tells you is, "Hey, one of the big, "big phrases that people use in addition to 'iPod car' is iTrip." And this site doesn't use that word once. Question. Yeah. Mm-hmm. Absolutely. So the question is, "How do you get around "the fact that, you know, maybe you don't-- maybe you don't sell the iTrip?" The fact is, people make, you know, there are affiliate programs for the iTrip. So what you can see right here is iTrip is queried, like, 240,000 a month in April. And if you were to do the query "iTrip," um, there's Griffin, there's all this stuff. I happen to know the guy that's number one here, right? And it's one guy in New York. So it's not that iTrip is some hugely competitive phrase that you could never show up for. They could at least put a page about the iTrip and start to show up for some of those searches. Now, I don't want to get too hung up. We've all--we've spent almost ten minutes. So let's, uh, look at a couple other sites. Um... is the owner of fazalfurniture.com in the room? Good. Good. I was hoping they wouldn't be, um, because I have to give a little bit of tough love to fazalfurniture. So here's what the site looks like. Uh, notice there's a lot of bolding on this site. If you were a visitor and you landed on this page, would you be comfortable with this amount of bolding? Uh, read a few of the words. It's almost like somebody's not necessarily a native speaker. This is the very first thing a visitor's going to see. They're--that's going to be their impression of your site. So what's interesting is this site is only about six pages deep. There's a front page and then there's a details form for each one of these phrases. And what's kind of interesting is you click to get more details on this, say, Indonesian chair, so you get a little picture, the picture never gets bigger, right? "Oh, I want more details. Yeah, show me a bigger picture." No, the picture never changes. It stays the same size. So there's all kinds of ways that you could entice people to order, and they're not necessarily going to get enough information to feel comfortable. There's something else you need to think about, which is the fit and finish of your Website. And we'll go through a couple examples of that. Does anybody see anything that either gives them a huge amount of confidence in buying from this site or maybe a little bit of pause where they'd worry? Yeah, the co-- yeah, the contact order, you know, is--it turns out it's just a Web form. See this Rattan furniture? If you click to view details, uh, they keep spelling Rattan a different way, right? And if I were buying for a motel, I'd be, like, "Well, don't you know how to spell the type of furniture that you're selling?" Either you do, in which case you're deliberately misspelling, which looks bad, or you don't. And if you don't know how to spell the ingredients of your, uh, your furniture, that's a pretty bad sign as well. So this is the sort of thing where a-- See, right here they're calling it "rottan" furniture, right? And it may very well be rotten furniture. We don't know. But--yeah, go for it. Can we, uh, get the other lap up if... White: Looks like your background image it's using, the nice--really nice word grain is, um, 1 megabyte. 1,000 K. Cutts: Right. White: It comes in... Cutts: No wonder my page load is failing. A one-- yeah, a one-megabyte download-- Okay, so we have pretty good connectivity here. My in-laws live in Omaha. So, like, every Christmas, I have to go to Omaha. And I try to load the Google home page. And for a long time, they were on dial-up. How long do you think it took me to load the Google home page, which is the most spartan, austere page, like, on the Web? Like 12 seconds on a modem. If you want to load a one-megabyte background, you can pretty much start at, you know--at the beginning of this session and then come back at the end of this session. And then by the time you're done, you're like, "Oh, this site sort of sucks a little bit," right? So pay attention to the fit and finish. Pay attention to how it looks for the users, because if the users like it, you'll get more conversions. And it can be all kinds of little things. For example, "about us." Oh, good, I can find out more about this person. Click. Click. Click. Crap, It's not a link, right? So on one hand, this guy's done a pretty good job. He's bought some sort of template or something like that. So it's got a "contact us" and "about us" and "home page." But he didn't have an "about us" page so he just left that text lying there. And that's not necessarily very helpful. So pay attention to the fit and finish. Pay attention to your images, how fast things download. Please don't call your furniture "rottan" if you're trying to spell it. And think about how the user's going to perceive it. There's a lot of key words here, right? He--How many times does this guy use the word "furniture"? A lot, right? It's not the case that if you repeat the word "furniture" 100 times, you will do twice as well in Google as if you say the word "furniture" 50 times. I promise you. We're pretty smart about that. After the first two or three times, it ain't going to help to say it 85 more times. So he doesn't need to say, "Furniture, furniture, furniture, furniture, furniture." If someone talked like this, they'd be the most annoying person in the world, right? So you may work with SEOs. You may be an SEO. And you may get this question, "Well, what should my keyword density be?" Don't worry about your keyword density. If you have the word on your page two, three times, Google knows about that word. Think about reading it out loud. If it sounds like something normal if someone were reading it out loud, that's pretty good. It doesn't need to be like you're a bot saying the word "furniture" over and over and over again. And there's a few interesting things here. He focuses on the word "furniture," but he could probably think about other synonyms, right? Go to the Google keyword tool. There's lots of keyword tools around the Web. What are some other words for "furniture" that he could use instead of "furniture"? That's the sort of thing that users might type in. Okay, so here's a little interlude. Savetheinternet.com is dedicated to net neutrality, Internet heroes and villains. It's, uh, hosted by Free Press. This is a cool site. This is a site which is about issues. This is a site about spreading the word on net neutrality, which, personally, I think is a pretty cool idea, except... See this, uh-- see, I was on savetheinternet. I'm going to go check out their blog. So I go and I try to access blog/wp-content/uploads/authors /pleddle.html. What do you think-- assuming the Wi-Fi holds-- what do you think I'm going to land on if I'm looking at this site's blog? Am I going to learn more about net neutrality? Am I going to learn about freedom of the press? If you look down here, now it's waiting for healthshoponline. Why is a site about net neutrality linking me to healthshoponline? It turns out this site has been hacked. And if you visit the blog, you'll be offered ***, okay? So the point that I want to get across-- and here it comes. Canadian Pharmacy. Number-one Internet online drug store, right? You guys, as developers, are due for, like, the most concentrated attack of hackers that you would believe. It used to be people would get individual PCs and hack their botnets. And as PCs get better, as more people roll out things like Windows 7 or switch to a Mac or use different browsers that are more hardened, the clients are-- are getting pretty good. It's hard to hack one of these bots-- one of these PCs. So now people are going to be attacking your Web server. So you have to keep your server patched. This savetheinternet, I've been to it before. I was like, "Oh, what are they doing selling pills?" And it turns out, you know, if you click on "blog," it says something like, uh, "We've been hacked. We're trying to figure out what's going on." And yet the hack is still live, right? So as soon as you find out that you've been hacked, take it all down. Go ahead, restore from backups. You have to apply your server stuff really frequently. So stuff like WordPress, for example, comes out with relatively frequent updates, and it's relatively easy. So now they've gotten the blog back, but they don't realize that they're still selling pills. So if you have a site, one thing that I would absolutely recommend, just run it as a daily search, or set it up in, you know, uh, just a cron job or something. Do site:yourdomain.com ***. Or site:yourdomain.com pills, or online gambling, or XENICAL, or Hydrocodone. You know, pick your favorite debt consolidation, mortgages, spammy kind of query, because if you're hacked, then you'll find out about it really soon, and you'll be able to fix it before it starts to effect your search rankings. Questions about that? Make sense? Yeah. Yeah. Yep. Yep. Yeah. So the--the question is, "If you use WordPress, you know, it's pretty easy to get hacked." And the fact is, WordPress has gotten much, much better. You know, they're including one-time keys, all sorts of nonces, things that protect you. So 2.7.1, 2.7.0 are a lot less likely to be hacked. But if you use a stupid password-- like, suppose their admin password was "savetheinternet," people could still guess it. People can still hack it. We're actually in the process-- there's a--there's a site, google.com/webmasters, and whenever we see your site get hacked, we will often try to send you a message and say, "Hey, we found the following text on your page." Now, we can't detect that you're hacked every single time. But one thing that we've done is we've scanned for obsolete versions of software. And we've said, "Okay, if you're running WordPress 2.1.0, you're basically naked on the Internet," right? If you're running a version of WordPress, you know, 2.2.1, you're still naked on the Internet. And it's not WordPress. It's Joomla!, it's, uh, you know, all sorts of different-- Drupal-- all these packages have had holes at different times. So Google will try to give you a warning, but we don't guarantee you that we'll warn you. Yeah. [man speaking indistinctly] Yeah. [man speaking indistinctly] Yeah. Yeah. So we can take it offline a little bit. But the basic idea is, if you notice that you've been hacked, take your site down, restore from backups, make sure that you're as secure as possible. And then you can do what's known as a reconsideration request, at least for Google. And that says, "Hey, we're okay. We've cleaned up the hack." We also have a Google Webmaster forum. And that's linked to from google.com/webmasters. So that's a couple ways you can let Google know that things are, uh-- are fixed. So this one's kind of an interesting site. This is *** InSite at ucf-- ucsf.edu. Pretty cool site. Um, I noticed a-a couple interesting things. Number one, they're really not using very much of this page. So if you want to see what's going on, you have to go all the way down here to see there's--there's basics. There's different sort of things. The other thing-- so I'll go ahead and use the rest of this space by making the font bigger. Um, check out this one. Recommendations for Use of Antiretroviral Drugs in Pregnant ***-1-Infected Women for Maternal Health and Interventions to Reduce Perinatal ***-1 Transmission in the United States. What is that about? I have no idea. I'm not an *** expert. Is this the very first thing that you want your users to see? Maybe it is. Maybe this public-health site really wants to highlight what are the important medical papers coming out. But if you're taking the time-- And notice this says, "Editors' Picks." So someone at UCSF has decided that that's an important paper. If you're taking the time to decide that's an important paper, include, like, a paragraph of beginner language. What does this mean? "Oh, if you are a new mother, "it will help you if you do the following steps. Click here to read more." Right? So what's really interesting is if you go down to the bottom, there's a section called "Basics." And I'll just tell you, you know, it's possible that you can go to Google, and you can, uh-- if you go to Google and--and stand in the lobby, you can see queries streaming by. And if you ever do that, what you'll find is that typically, the sort of queries that show up are not *** transmission in maternal blah blah blah blah blah. Here's what people type. Make it really big. [hums] You'd be surprised how many people type those queries, right? "Aah! "I went to the doctor. "The doctor used this really long phrase. "And I don't know if this is a serious disease or not, you know?" So people who come to Google type in things like this. You'd be amazed. Something like 25% of all the queries we get in a given month are unique. And it's because it goes straight down deep into the long tail. And they use beginner language. They are not using language like, uh... You know, here's the Basics section. "What are *** and AIDS?" "Should I get tested for ***?" How about this question. "I just tested positive -- now what?" "I just tested negative -- now what?" If you are designing this purely for users, I would probably put this section front and center, right? And move the research papers a little bit further down, 'cause if I'm a user, I have to click a lot further to find this. I just happened to see the Basics section. This is the part that's going to be of a lot more interest to your average user. So, you know, when people come to Google and they start typing this, this is actually a pretty good match for this sort of thing. So there's a ton of good resources here, but you might think about rejuggling a little bit so it's a little more crawlable, a little more, um, understandable. So we call this, like, the disconnect in jargon. Any of you guys do industry sites with heavy-- yeah, like, technical language? Think about what the user is going to type, right? That guy who was doing iPod car integration, he's completely ignoring the 250,000 people last month who said, "iTrip." He doesn't even have a mention of that product on his page. In the same way, this site, while very good-- in fact, very crawlable, right?-- sort of doesn't necessarily match with the terms that users are going to use. And you want to think about the first impression you have whenever you land on a page. Does that make sense? Cool. All right. Let's look at another site. Ah. What do you think? This is what my team sees all day, every day, all day long. Uh, blogsense.biz. Uh, post number 1,276. So he's done 1,275 previous blog posts just like this. "Radmaxx kicker fx, phantom electric scooter, used electric scooter, schwin, new." So don't do this kind of stuff, right? If you ever get somebody coming to you and saying, "Hey, uh, we've got an idea to generate "a lot of envelope pages, hallway pages, doorway pages, shadow domains, whatever you want to call them." A lot of things that-- "Oh, it's okay. there's no spam on your domain. "We put the spam on other places, and then we sort of funnel them in your direction." Users hate this kind of stuff. Check out the previous post. "Aamaryland Child Support Check myIowa," right? It's pretty much complete gibberish. And I hope that programmers in the audience are not thinking, "How could I generate a page like this?" I'll tell you right now, they use things like Hidden Markov Models. They scrape search results. They try to find out search results-- search queries that people have done. But the nice thing is this stuff is pretty detectable. Uh, it's relatively rare to see pages like this on the Web these days. Or they exist on the Web, but you're less likely to find them through search engines. Notice that even the spammer's getting spammed. He's got a WordPress template, and it says, "Download movies, mortgage calculator gadgets," and he's got little links down at the bottom. So a lot of this stuff is just completely awful links. The reason why I wanted to show you this one is a lot of the times, people come to me-- And I'm guessing almost everyone in this audience is going to be, "Okay, should I break my articles "into multiple pages? "You know, what if I have a one-page thing "where I can print the entire article? Is that going to be duplicate content?" It's amazing to me how many white hat people come, and they're worried about something that's, like, you know-- I can understand why they're worried, but they really don't need to worry nearly as much as these guys. The black hat guys don't skirt the line. They go--boosh-- right through that barrier and break through it and go as fast as they can toward the keyword stuff and sort of spam stuff. So, uh, it's natural that people worry about different things. But most of the times, you don't need to stress that much. Okay, let's look at another one-- Comfortfeetshop.com. What do you like, and what do you dislike? Get a little audience participation. What do you think is good about this page? There's some pictures. Someone else was saying something else? Brands-- actual brand names, yes. Mephisto. I've heard of that. ECCO. I've heard of that. Okay, what do you dislike? Brands. There's--there's a lot of brands here. Now, and you said-- Yes, yeah, this-- things need to be centered. You know, it's kind of weird that it's left-justified here. and centered here. That's a little strange. The text is really weird. Look at this. Oh, yeah, I would like to buy some Cole Haan. I can't buy it. It won't let me buy it. Please, give me some Cole Haan, right? So number one, that text looks a little-- sub-optimal is a polite way to say it. And number two, it's not even clickable. So if you do want to get some Cole Haan, you can't click on it and get to it. It turns out the secret is... that the individual images are clickable. So there's a really good principle of Web design, which is, "Keep it simple, stupid," right? And that is users are just going to click on random things all over the place, everywhere. And so imagine--imagine if you have an RSS icon, and you have RSS feed as anchor text. Make both of those clickable, right? In the same way you want to make the image clickable and the text clickable. Don't put it, you know, a lot of the times, you can learn something by saying, "You know what? Let's get a regular user." I swear, probably the best $20 you can spend is grab a regular person off the street, put them in front of your Web page, sit on your hands-- you're not allowed to say anything-- and watch them try to buy, right? And so if they tried to click on this-- Of course, the Wi-Fi's gone. If, uh--if they try to click on a--on text and they can't get to the brands that they want to buy, they get really irritated by that. While we're waiting for this page to load, what do you notice about the URL? It is meaningless. There's no way a user can remember this particular URL. In fact, did you notice it has an exclamation point in it? You don't see URLs with exclamation points in them very often. And, in fact, if you looked at the home page, the, uh--see if I can hit the back button without hitting reload-- the URL was this, right? So there's clearly a content-management system-- a CMS--that is loading the home page, which is homepage.template. I like to call this, "When you're fighting with your CMS." You shouldn't have to fight with your CMS. Whenever you have, "Oh, yeah, go to the product URL, it's just gez!mep yada yada yada," users are not going to remember that as much. Search engines are barely going to be able to crawl that. So if you can go to comfortfeetshop.com, see, even then, it still redirects me to the deep URL. And so if you can, it's much better if you can stick with short URLs, if they can be very memorable, if you don't have to fight with your CMS. And the thing is, it's 2009. We shouldn't be fighting with CMSes at this point. If there's a CMS that isn't built to be search engine friendly, consider going to a different CMS, right? Be that voice that says, "You know what? "If the CMS hasn't even thought "about what search engines might do, "maybe you need to switch to something a little bit different." Does that make sense? Yeah, question. All right, so we'll do this-- we'll do one go around, which is this power-user question, which is, suppose I've got two links on a page. They both go to the same place. Should I put a rel="nofollow" on one of those links? So first let's back up and talk about what nofollow is. Nofollow is a simple attribute that you can put on links that says, "you know what? Don't necessarily float page rank. Don't necessarily float anchor text across this link." And my short answer is no. In general, whenever you're linking around within your site, don't use nofollow. Just go ahead and link to whatever stuff. And the site architecture that you decide to use decides where the page rank flows within your site. That's why if you have a treelike structure and you have the most important stuff first, it'll get the most page rank, because most people link to the root of your page. So you remember the *** InSite one, right? They had the medical articles up front, and then the basics were down here, buried two or three mouse clicks deep. That's a good argument to swap those two. You'll get more page rank for the Basics and the deep medical articles that less people are interested are further down. So a good way to think about it-- and it works for search engines as well as for users-- is how many mouse clicks away am I? How many mouse clicks does it take to buy your iPod car integration? How many mouse clicks does it take to buy your women's sandals? So that is almost kind of a good proxy of page rank, because page rank is like a random surfer. Whenever we find a page, we extract all the outgoing links. We take the out degree. We divide your page rank by the out degree. And page rank more or less flows in that proportion to all those pages. So the closer you are to the root page in terms of links, the more page rank your page will have. So if you have to clink 17 links to find, you know, a particular page, that means Googlebot has to click 17 links to get to that particular page. So the amount of page rank will be much, much less. That's why you should think about putting your most important stuff relatively close to the root. Now, I'm talking in a virtual sense. Imagine if you click one link, and it gets you three directories deep. Is that a problem? On Google, it's not. We only look at, "Oh, there was one link, "and it went, you know, three directories deep within the site." That's totally fine. It's just one link. There are some search engines who probably look at the number of depth of the directory. So if you can make it relatively close to the root on the number-- the number of subdirectories, that's not a bad practice either. Then you might do a little bit better on Yahoo! You might do a little bit better on Microsoft. But on Google, you don't need to worry about that. So the question, pulling back, is, should I put nofollow on one of these links? No. Let all the page rank flow. The only rank time I'd really use nofollow on your own internal links where you say, "No, don't flow any page rank," is maybe a login page, right? Think about Expedia. If I go to Expedia and there's a login page, what good is it if Googlebot crawls that login page? We don't have a credit card. Googlebot's not going to fly to Vegas for the weekend. So you don't need us to, you know, go and visit your login page. So that would be a good one to put nofollow on. Don't--Don't bother to waste the page rank on, you know, uh, that sort of stuff happening. Mm-hmm. Tell you what... maybe--maybe take it offline and catch it after. Okay, now we reach the very active participation part. I want everybody to close your eyes. I still see open eyes. No, seriously, close your eyes. Close your eyes. While I have your eyes closed, I'm going to bring up a site. Keep your eyes closed. Okay, open your eyes. Snap judgment. What's this site about? What does this site do? No idea. Closebys. "Closeby services meet Close friends to share the small things." What the heck does that mean? I have no idea. Okay, well, let's dig deeper. "3 cents to 10 cents/ACElet upto $110/year." What does that mean? I have no idea. Okay. Um, "Everyone qualifies for 3 cents." Yes! I've made 3 cents. What does that mean? I have no idea. Okay, so let's start digging deeper. Let's start clicking around within the site. "Hair Salons. Auto. Maintenance. Home." Now, if you're looking in the bottom left, every single one of them leads to the same URL. They all lead to closebys.com/services. Okay, so I'm getting confused here. "Maintenance. Health & Fitness. Travel." Okay, let's click on Travel. If someone can't land on your site and know what it's about in a second or two, you've lost probably half your buyers or half of your conversions or half of whatever it is you're trying to optimize, right? So this is a good example where you get that person, and you say, "Okay, I will give you $20 to use my site." And probably the first five people to use that site are like, "What--What is it?", right? Oh, we sell ACElets. What are ACElets? ACE is an announcement, coupon, or event sent to you by your local business. Okay, so now we start to get to it. You register for ACElets, and then you can get these sort of offers sent to you, and they can be coupons. So that can be pretty useful. Um... start to look into it a little bit more. Services pay per customer. It's a little bit opaque, right? What does Closebysocial have to do with businesses allocated, right? And so this is another one where you want to sort of think about the fit and finish of your site. I-I--We were doing a practice session at Google. And I was saying, "Okay, close your eyes. Open your eyes. And what does this site do?" And somebody was, like, "They sell guns!" "What? Why do they sell guns?" "'Cause there's an NRA show." "NRA. Maybe the-- Oh, that's the restaurant. Okay." They seriously thought, and I seriously thought until I moused over it, that this was about selling guns, right? So think about the overall fit and finish. Um, and here's a simple example. "Upto" is two words, right? That's like the fifth word that you see whenever you start reading the page. So think about the little things. Think about the fit and finish. Think about--There's actually a pixel missing here, which you might or might not be able to see. And so how--how professional do these things look? Um, as you click around, it's actually kind of an interesting idea. If someone wants to save money, they can sign up and get these announcements, coupons, and events, stuff like that. So it can be pretty handy. But you're not doing a very good job of communicating it to people. Now, I've--I've been a little sarcastic. I've been a little mean. So what's good about this site? Anybody want to chime in? What do they think? You won't--You won't fall asleep looking at it. That's true. It looks a little like a MySpace page. Got the glitter. You know, that would sort of complete the look a little bit. Um, I think this is kind of interesting too. This is a banner ad on Closebys for Closeby services. And if you click on it, you go back to Closebys, right? Fit and finish. It's not all about search engines. It's also about your users. It's also about your visitors. Now, it's not completely the same URL. In fact, if you look at it, it's actually closebys/index.html. And that brings me to a topic that I wanted to mention. You would be amazed how many people have got closebys.com/index.html, index.php, noindex.html, www.closebys, non-www closebys. So one of the big pieces of advice that I give is look at the internal linking on your site and be consistent. You would be amazed-- literally amazed-- at how many sites have a www and a non-www version of their Website. Now, why would that matter to you? Greg, Brian, why would it be a bad thing if I had two different versions of my Web page? Grothaus: Well, one-- for one reason, if people are going to your Website, uh, they'll go to one page or the other. And as a result, they'll--they're sharing that Website with their friends. They'll send you one or the other. And all of your links on the Web will end up being split between those two different versions of the site. Uh, Google's pretty good at noticing this. But sometimes when we see that maybe these two pages look different, maybe we'll crawl one version and crawl another version at a different time. And you have changed your site dramatically in between them. And so we're a little--We can get a little bit confused. Are these the same page, or are they not? And maybe we'll split your link juice between the two pages and not give you full credit for all your links to one single site, which is what you really want. Cutts: Absolutely. We normally do a good job. We normally can say these are the same. But imagine if this is a dynamic page, right? What if right up at the top of the page, you say "the current date is" with a time stamp? And then we visit www and non-www, and the pages look a little different. We might not correctly glom those together. So I've actually seen a www have a page rank 6 and a non-www of the page have a page rank 4. So they're splitting their page rank in between those pages. Whenever you can, it's really nice if you can kind of unify them and bring them together a little bit. Questions about that? Yeah. Oh, speak up. Shout it out. White: Can you go to the mic, perhaps? Cutts: In fact, come to the-- come to the microphone. Guy's like, "I didn't want to be embarrassed like this. Come on." man: Um, does it help to have, like, a redirecting, uh, application or something like that for those type of domains? Like, doing 301 redirects or something? Cutts: Yeah, absolutely. Um, it helps to have 301s internally and externally. So Google is a very large Website. A lot of people come to Google. And so a lot of people register domains that sound like Google or variants of Google. For example, porngoogle, googlesex, googlingfordollars, right? All these people, they are like, "Oh, I'm going to make money off the Google domain name," right? And so they register something something Google. And then we have to go and say, "Hey, you didn't invent the word Google. "Can we have googleporn or googlesex or whatever back?" And they're like, "Okay, fine. Here you go." Now we have this domain, googlesex.com. What are we going to do with that? We're not going to have an unsafe version of Google that's, like, only ***, right? Matt's house of ***. Do all your searches here and get all the sort of weird, fetish stuff you want. We're not going to do that. So what we do instead is we do a 301 redirect. And that's a permanent redirect. And 301 just refers to the HTTP status code. You can pretend to be a browser using Telnet. So if you Telnet to port 80 and say, "Get slash," that's saying, "Give me the root of this page on your Web server." And the--the response code is 200, okay, if it's a normal page. If it's moved, it's usual 3-something. If it's 301, it's permanent. If it's 302, it's temporary. So imagine that you've just moved your Website for, you know, a few weeks. That's the perfect time to use a 302, if you're going to be coming back to it. Um, a lot of people who do banner ads will do 302s, things like that. But if you have moved your site-- if you have truly moved it from sitea.com to siteb.com and you're never planning on going back, that's when you can use a 301. And then all the page rank flows, and everybody's happy, and your rankings are preserved and all that sort of stuff. And you can do an internal 301. That means if you go to the non-www, you get a 301 to the www. So that's really nice. You're not linking to it. You're just--voosh--immediately taken in there. And a lot of sites like Slashdot. If you go to www.slashdot, it'll probably redirect you to just slashdot. And that's pretty smart. Then people can't really get mixed up. If you try to link to it and you visit it, you'll get redirected to the non-www and then you end up usually linking to the non-www. So that's a great way to take care of those sort of duplicate issues. White: I'll add to that too. Cutts: Yeah. White: A number of Web hosts have already taken care of that for you. On my Web host, um, I found that when I typed a non-www-version domain name, it automatically redirected over or vice versa. So that's sometimes taken care of for you. I want to make a plug, too, for Live HTTP Headers extension in Firefox. Does anyone use that? Raise your hand. Yeah. That's a-- That's a nice one. Briefly, it lets you see the requests and the responses between the browser and the server. In--In effect, you are kind of-- I don't know a better way to say it-- but wiretapping your own, uh, connection between the servers. So you can actually see, um, what the response codes that come through. You can do-- Use Matt's version. Telnet. You can use Live HTTP Headers and make sure that 301 is happening from www to non-www or vice versa. So, yeah, that's something you can add. Cutts: It's a really useful tool. You would be amazed at how many sites return very strange stuff, um, you know, in the headers. And so if you're looking at those headers, you're realizing, "Oh, something strange is going on." And you can do it yourself. You can go down to the wire and Telnet and all that stuff. But it's much easier if it's just a Firefox add-on. So it's a lot of fun. One more plug, and I don't want to-- We are at a Google conference, so I don't want to be, you know, too salesy, but there's google.com/webmasters. That's the Webmaster console. If you haven't registered, it's not negligence if you don't use it, but it's a really, really, really useful tool. For example, if you say, "Okay, I own mattcutts.com," we have an option where you can say, "What's my preference, www.mattcutts.com or non-www?" We also show you errors that we found when Googlebot crawled. We even showed you how long the latency is for Google to fetch pages from your site. So if you're having server issues, you can sometimes diagnose it. So we'll show you all kinds of really useful information. And the ability to set that sort of www or non-www is very helpful in that kind of case. Question. Yeah. At the microphone, yeah. man: Um, yeah, the, um, question before about variants in-- Um, for example, I use Analytics extensively. Cutts: Mm-hmm. man: And I'll get a return for the URL-- the URL plus, uh, index.htm-- um, several variants. And I don't think that's a question-- I think I'm consistent within the Website. Cutts: Mm-hmm. man: But, um, which kind of contradicts what you just said. Cutts: Oh, no, you always want to be consistent if you can. man: Yeah, but I mean, those external accesses, how did they, um, decide, you know, to access my Website when I was consistent internally, I guess, is the question. Cutts: So most of the time, it's just people are not that skilled. Like, they're going to link to Slashdot, and so they just assume it's www.slashdot, even though Slashdot wanted to be without the www. So a lot of the times, if people-- You would be amazed at how many people will link to you with broken stuff. In fact, you know, to plug the Google Webmaster console, we will show you all the backlinks to your site. And we'll also show you all the 404s on your site. So one of the best tips that I always give is, if you would like free links-- and most people on the Web would like free links-- you can go to this tool and you can say, "Who are the people who link to 404 pages?" And they're going to be very reputable people. There was, uh-- ZDNet had linked to my site, and they'd linked to a page that didn't exist with, like, two trailing slashes. And so if you contact that reporter and you say, "Hey, uh, you know, double slashes on the Internet are weird, just link to the normal page," hey, free link. Now it's not a 404 page. It links right to me. The fact is, anybody can link to anybody else on the Web. So we don't penalize people if a spammer links to you. The nice thing is, if someone links to you that you normally link to correctly, but they don't always link to you correctly, by being consistent with your internal linking, you try to prevent at least the worst of stuff from happening. Good question. Yeah. Another question. woman: Um, we, uh, work at the University of California, and we're not allowed to use some Google tools because of an indemnity clause. And I'm wondering if there's an indemnity clause associated with the Webmasters. Cutts: Uh, that's an excellent question. Catch me and get my contact info afterwards, and I'll--I'll try and look it up for you. Yeah, this is one thing that, um-- I talk to a lot of government Webmasters, right? 'Cause they're like, "Okay, we would like to find out "what the broken links are on our site, but, you know, bureaucracy can get a little overwhelming." And so sometimes people literally just go rogue. They're like, "I'm just going to prove "that I own this Website. "I'm going to edit this meta tag. "No one needs to know. "Okay, now I'll log out. And I have the information I need." And I'm not condoning that. I'm not sanctioning that. I'm just saying a lot of people, turns out, they do that. Uh, but everybody has different, you know, restrictions on when they are or aren't allowed to do it. Um, I'll have to check on whether we have some sort of indemnity clause on--on Webmaster central. Good question. Another question. Yeah. man: Hi. Um, interesting that you're from Omaha. We just, uh--we sold our company a couple of years ago, NetShops. Um, and because of non-competes, we--we really couldn't sell much of anything online. So we started that racy.com site. And I understand why you didn't put it up. But I-I'm very, very curious, are we starting from a hole because we're trying to-- Uh, it's not pornographic, but it's on the edge. And I'm just wondering what, uh--what the take-- is it--is they're also history of the domain name that is bringing along problems? And how, uh-- how can we resolve-- What--What should we do in this situation? Cutts: Absolutely. Very fun question. So we took a bunch of submissions. Um, one of the submissions was racy.com, which I have no problem with. I put it up, you know, in our internal training session. We were looking at it. There's some pretty explicit stuff. In fact, last night I was trying to find some part of the site that would be safe, you know, to project. And I was like, "This is a little--" It's not radioactive. It's just, like, there will always be a couple objectionable words, so I decided to not put it up on the projector. Uh, but as I recall, you had good URL structure. So let me turn that into a more general question. I just bought a domain, and I'm worried the previous owner might've done something unsavory with it. It turns out, in your case, that didn't happen. But here's what you can do, go to archive.org and look at www.racy.com or example.com or whatever your site is over history. In fact, if you're thinking about buying a domain name-- and we have seen people buy domain names where a spammer took it, and they burned it to the ground, right? They just did everything. They spammed-- You remember the, uh... this--this--this guy-- Where's my spammer? There's my spammer. Blogsense.biz. A lot of the times, people will buy this domain, spam the hell out of it, do as much stuff as they can, and as soon as all the search engines have caught them and, like, deconstructed their spam house brick by brick and buried the bricks in different places and then salted the ground so that this domain does not rank well, right? Then they try to sell the domain, right? And a lot of the times, if your traffic is going like this and then the search engines find your spam and it goes like this, they'll take a picture when the traffic was like this. And they'll be like, "Oh, yeah, look at our rankings. They're really climbing." So the way to protect yourself against that is a couple things. Go to archive.org and search, and you can see the historical amount of information about the domain. You can see what the domain looked like three years ago or four years ago or five years ago. The other thing is, just do a search for the domain. I think I can do a search for racy.com, um... as a phrase, right? So racy.com is right at number one. That's a good sign. And you'll notice now you have dictionaries, that sort of stuff. If what you see is people complaining about your site, people saying, "Oh, I can't believe racy.com. Those guys are spammers," or something like that, then I would be worried. But in fact, I don't think that you necessarily have any issue with the previous owners having issues or something like that. Feel free to catch me afterwards, and we can talk about it more then. Okay, more questions from the audience? Yeah, come on up to the mic. I'm figuring you guys prefer a little more interactivity, but I'm pretty-- man: I have two questions. Cutts: Cool. man: Do you treat 500 responses different from 502s? Cutts: Mm-hmm. man: The second one is, um, images. Do you guy-- How--how would a-- Let's say we have a thumbnail version, and we have a full version. And we prefer you index the full version, but we're only showing the thumbnail on a particular article. You can see an example of that on ehow.com. Cutts: Yeah, absolutely. So 500 versus 502-- if you want to catch me afterwards-- I think that we do treat those separately, but I'm not 100% sure. Um, there's a lot of different HTTP status codes. So 300s typically stand for redirects. 200 usually means okay. You know, 400 is some sort of error. And then, you know, they mean different things. Um, typically whenever we're doing image searches-- so let's do an image search for flowers-- um, we will try to find the best picture that we can. And if there's a thumbnail, we will do our best to say, "Okay, let's find the one that's a little bit bigger," because users are typically a little more interested in the one that's bigger. One thing you can do is try to shunt a little more page rank to the larger image. That can help out quite a bit. Um, we can also talk about it offline too. Did you--Okay. Yeah. man: Hi. I had a question about a good site. If we could just look at a site that follows some of the best practices. And, uh--And I also had another conference where they talked about video being really important on a Website to boost, uh, ranking and also indexing. Cutts: Yeah, absolutely. So--And the funny thing is, uh, what was it, Closebys-- I may have lost it. Um, Closebys actually had a very high-production video, and they buried it, like, six layers deep. So, uh, a lot of people ask me, you know, "Hey, I want to get number-one search rankings." It turns out, sometimes it's easier to get number-one search rankings by making a great video or a great blog post or a great image or something that's newsworthy rather than just a pure Web page. Um, video, you would be amazed at the sort of-- you know, if you go to YouTube-- the amount of traffic that some videos can make. In fact, uh, if I were to search for, uh, Google Webmaster video channel-- One thing that we've been doing recently is not just recording-- not just doing blog posts, but doing videos. And they're fantastic, because you can take a Flip, right? And Flip video costs... 100 bucks, a couple hundred bucks, right? And you can--I mean, it's like the size of a cell phone. You can carry it with you everywhere. And so if you see something interesting at a conference, you just whip that out, and you can video it very easily. So there's no reason why you have to put a ton of work into a blog post. Sometimes you can do it a lot easier with a video. Uh, I was doing something where I was hacking a Nintendo Wiimote. And I was like, "Oh, this would look good with a video." So you just whip out the Flip, upload to YouTube. You're done in no time at all. So we've discovered that doing-- We've done something like 50 different videos. And, um, what we do is about every month or two, I take questions. So, you know, "How important are brands and ranking?" "Does anchor text carry through 301s?" I can answer that question in a minute and a half instead of 20 minutes or 40 minutes or an hour doing a blog post. And you can see, you know, 14,000 people have watched that first video. So two minutes to make, and then you can get a ton of views. Video is a very easy way, where, if you are willing, if you have the ability-- You have to worry about indemnity. You have to think, "Do I have permission to make these videos?" But if you do, you'd be amazed at how viral they can go. Um, one tip that I learned-- You notice every single one of these has me in a red shirt, right? So I sat down, and I answered 30 questions all on one time. And by the end of 30 days, everyb-- I was doing one of these videos a day. By the end, everybody was like, "Okay, no more of the red shirt. "Please burn the red shirt. "I'm so tired of this red shirt. Ah, I hate it." So I did another one recently. And I brought in eight different T-shirts. And I do about five questions, and then I'd switch to a different colored shirt. So, you know, these things are very fast to make. You can make 30 or 40 in a day. And it can be a way to rank really well. In fact, if you were to do a search for my name, um, you'd probably get my blog. You'd get, maybe, you know, something on Wikipedia or whatever. But you--There's something like five videos ranking for my name. So it can be a very easy way to rank. And there's not that many people that are thinking about making videos, so it can be a really big opportunity. Yeah, another question at the mic. man: Uh, well, our marketing people really like PDFs, and so they create them first. And then, um--and then I'm one of the people who turn them into HTML. And I keep telling them to do-- You know, "We need to do it the other way." Uh, could you speak about PDFs versus HTML and what's likely to be most searchable and et cetera? Cutts: Absolutely. So--and just before I go on, here's a couple videos. Qualities of a good site and the lightning round-- a bunch of questions. And that's ranking with, like, no links compared to--these guys might have hundreds of links. So videos can be a really big opportunity. Um, yes. PDFs. Another one I get is Flash, right? Um, I love Adobe. Adobe has done fantastic things. Photoshop is cool. The plug-in aspect of Photoshop was revolutionary. They do amazing things. Postscript, fantastic. Flash and PDF are both very, very cool. But they're not necessarily native to the Web. And so people can usually read PDFs, but not always. And you want it to be as easy as possible for people to buy, for people to learn, for people to do whatever it is you want them to do on your site. So think about that a little bit. I would totally agree if you can make Web pages and then go to PDF, that can be a lot more accessible. It's a lot easier to link to an HTML page. So HTML pages tend to attract more page rank a little bit faster, 'cause people have to link to PDF, and then they have to say it's a PDF link, all that sort of stuff. And just on Flash for a minute-- Flash is fantastic. But it's better to be decoration in the middle and not the entire site in Flash. If you can make static HTML links that Googlebot or users can click, that is so much better, because we can find a lot of individual pages, which gives you a lot of individual opportunities to rank. If it's just one big glob of Flash, that's harder to link to. Anybody able to think of another reason why you might not want your entire site to be in Flash? Right? I have a G1, and I have an iPhone. The iPhone does not handle Flash well. And who knows when that will change. So there's a very great yogurt site-- Pinkberry, right? And if you go to Pinkberry on an iPhone, you're stuck with a little Flash cube. You're like, "I want to know where I can get yogurt. Give me the yogurt." And you can't find out. There's another site, you know, Red Mango, and they do a better job. So, you know, think about Flash. Think about PDFs. If you can do it in a Web way, I would start with a Web page and then go to PDF. Feel free to say, "Yeah, Matt said so," or something like that if it carries any weight. Often it doesn't. Question. Yeah. man: I had a question. Can you explain how the pa-- Google decides to index a page or not? I've noticed the page number in our site up and down, even though the content has not changed. Cutts: Absolutely. Um, the short answer is Google basically uses page rank to decide which pages to crawl and how fast to crawl. So in the early days, you could almost think of it as sorting by page rank. Stanford.edu, CNN, "New York Times"-- tons of people link to these sites. And so we start crawling with those sites. And then you find all the links from"New York Times," and you go outward like that. So it's almost entirely-- at least in the beginning-- a matter of how many links there were and what the importance of those links were. And by the way, page rank is not just the number of links, right? Like, if ten people link to Greg's Web page and 15 people link to my Web page, you might be like, "Aha, I got 15 links." But if his ten links are "New York Times," CNN, "Reader's Digest," and my 15 links are my college buddies, he's going to have more page rank. So we're going to crawl his page first. And the same thing applies to refresh policies. So we're more likely to visit stanford.edu, you know, once a day to figure out that a page has changed than we are somebody that nobody ever links to. That said, we try to do a very good job of crawling as much of the Web as we can. Even if you don't have any links at all on your dot-com, we can often still discover your site. You can always submit your URL and things like that. But the more people that do know about you, the more we're often able to crawl and the more often we're able to crawl you. So typically, we can refresh our entire index on a weekly or, at worst case, a monthly sort of basis. Another question. Yeah. man: I just had a couple quick questions. Uh, one of them regarding Google image search. Where do you guys-- What--What text do you look for to actually index the image on? Is it just the ALT text and image tag, or maybe like title text, or do you actually take stuff previous and after it? Cutts: The short answer is we're willing to look wherever we can find useful data. So, I mean, in theory, you could look at the title of the URL, definitely stuff surrounding the page, absolutely the ALT text. Um, it's very hard to understand the actual content of images. Google just released on Google Labs something that lets you sort by image similarity. So you can start with the Eiffel Tower, and it will show you other pictures of the Eiffel Tower. We also have something where, on Google image search, you can say, "Show me faces." So you can do a search like "Paris." And then click "show me faces," and you'll only get Paris Hilton. So we do a little bit of image understanding, but most of the time, it's text around or within the metadata of the image, that sort of stuff. The more descriptive stuff you can have, instead of, like, if you call your .jpg dsc129.jpg, that's not as helpful as flower.jpg. So think about all that stuff and try to incorporate the data. man: And if you throw an image tag in an H1 tag, will it get more authority like the ALT tags do? Cutts: This is a common question. H1, you know, is bigger on the page than H2, which is bigger than H3. And so a lot of SEOs are really, like, "Okay, everything's going to be in H1." And it's kind of nice we have this page up, right? It's the same attitude. I'm going to say the word furniture 40,000 times. I'm really about furniture. If you type in furniture, you have to return me, Google. And we don't buy that. So if you put your entire page in H1, uh, we're not going to give it the same weight that we'd give, like, two or three words. So we try to be pretty, you know, intelligent about that. I would just do whatever is natural. You know, it's natural to have an H1, have a few sections, have a few words, and have an H2. I wouldn't try to use CSS to make the whole page H1, because we do a pretty good job of detecting that. Yeah. Another question. man: Yeah. So I-I guess what I'm hearing from you, that if you, say, have a page-- you have, like, an index.php, and you give it query parameters like page=something, that's worse than if you had separate pages, you know, like, you know, whatever, index.php and, you know, item, you know, or, like, you know, iTrip or whatever .php, stuff like that. Cutts: Yeah. So that's a fantastic question. Imagine this page is index.php and val1 or, you know, item1=val1 and item2=val2 and you go on and on and on like that. It used to be that Google would only crawl URLs if there were, like, one or two parameters in the URL. We continue to get better. So you can have 15 parameters in the URL, and we can still crawl that. Like, if you looked at some of the weird URL structure we had with exclamation points in it, we'll still crawl that. But in general, for most search engines, if you can make something that's relatively static, instead of having 15 parameters, it's easier to crawl, because if you think about it, you know, if you have 15 parameters, usually at least a few of those parameters do nothing. So if you take that parameter out, you get the same URL. And now you've got weird duplicate content issues where the same content is presented on two different URLs. So as much as you can reduce those extra parameters if you don't need them, and, you know, if I were looking back at this guy, rather than having all these--whoops-- rather than having all these weird terms, again, if it could be /audi/s4/2006, that's so much more understandable for a user. So I think that can really be useful. It can-- you know, it can help out. We do give a little bit of weight for keywords in the URL. It's not like that's the secret. You put a keyword in the URL and now I rank number one. But every little bit helps. So if you can have a few things that make sense for users, where they can see the URL and get rid of these sort of multiple parameters, it does help a little bit. And I think we're pretty close to the end of the session. How about one last question. man: We put a database online that was unique information. We just did it. And I know the guidelines about length and number of links, but are those just guidelines? Can we break that in a situation that requires it? Cutts: Very good question. So if you look at Google's guidelines, there's one thing that says, "Avoid having over 100 links." And that's not in our spam guidelines. We're not going to penalize you if you have over 100 links. That's in a different section. The reason why we said that originally is because we used to only crawl 101 kilobytes. And so we didn't want people having a four-megabyte file. Remember, when you have a one-megabyte image download, that can be really kind of a pain. So originally, we recommended 100 links just so that people wouldn't have their pages be too big. Now we crawl a lot more-- We'll save a lot more than 101 kilobytes of your page. So that guideline is getting a little stale. You know, having up to hundreds of links on your page is okay. What you really don't want is a ton of really spammy links. You know, the sort of thing like, uh, the blogsearch.biz guy, like this. If you had 100 links like this, that would get really, really annoying. But if you have just normal links, that I wouldn't necessarily worry that much about. Okay, so thanks very much. If there's people who have questions, we'll stick around for a little while afterwards. Thanks. [applause]