SMX West 2012: Real Answers For Technical SEO Problems
Moderator: Vanessa Fox, Contributing Editor, Search Engine Land (@vanessafox)
Speakers:
Adam Audette, President, RKG (@audette)
Jonathan Hochman, Founder, Hochman Consultants
Maile Ohye, Senior Developer Programs Engineer, Google Inc. (@maileohye)
The session’s tags are #smx #22C. Vanessa says that audience questions will be a big part of this session, both through the online submit form and with the microphone going around. Intros first.
Jonathan’s consulting company builds websites and his startup does security. Maile has been at Google since 2005/6 working with search and tools team. Adam’s integrated agency does PPC and SEO. Vanessa worded at Google a long time ago and wrote “Marketing in the Age of Google.”
Rick Santorum has a Google problem. Some people were angry and they used their Google power to sully his SERP. The result by Dan Savage was ranked first for a long time. As of today it doesn’t show up. A SELand article asked if it was Panda, or a change to the valuing of links. How did the result drop? When you see a page completely drop out of the index like that, we have a tendency to think it’s a technical problem. In reality it’s gotten caught in the porn filter, according to Google’s statement. Vanessa says she’ll be looking into it further, as one audience member says that the result still doesn’t show up with Safe Search turned off.
Maile takes the podium to talk about multilingual sites and rel=”alternate” hreflang. A lot of sites are expanding internationally and with variations of languages. She’ll highlight 3 issues:
New tools are helping to remedy this issues.
rel=”alternate” hreflang lets you specify language or both language and region. Specify all alternates on each page. It can be on same or different domain.
Benefits of alternate hreflang:
- Helps consolidate several indexing signals.
- Displays local/targeted URL to searchers.
- May not be treated by other major search engines in the same manner.
rel=”canonical”:
- Signals preferred version
- Consolidates indexing and linking dsignals
- Displays canonical version url in search results.
- Also supported by Bing and Yahoo!
Use case: translated templates
- Often UGC site (forum or profile) whiere main content written in only one language
- Template availabe in multiple languages
- Can use rel-alternate-hreflang with rel=”canonical”
- A webmaster can say that the canonical version is es.example.com/javier-lopez because this is the site that Javier created with the content in the language he wants.
- The canonical version is indexed and is responsible for Title and snippet displayed.
- Indexing signals consolidated to canonical version.
- At search time, relevant language version of URL displayed
- www.example.com: Pretty item full of color (dollar sign)
- en-gb.example.com: Pretty item full of colour (pound sign)
- Canonical version indexed, responsible for title and snippet displayed
- At search time, canonical version of URL displayed along with regional relevant snippet
Use case: full translations (car, coche, auto!)
- rel-alternate-hreflang works on full translations
- Don’t use rel=”canonical”
- Several indexing signals consolidated
- Individual pages indexed, allows different langs for titles and snippets
- Targeted URL to serve searchers
Benefits:
- More targeted URL for searchers
- Helps discovery of new URLs
- Google better understands your site
Best practices:
More info is at http://goo.gl/0JY9z
Vanessa asks how many use canonical tag on pages – lots. Vanessa thinks the tag should be used on all pages. You never know if a tracking code will be added to a URL, even if not by you. That accounts for all possibilities. Bing is a weird case because they said they don’t want self-referencing canonical tags. She thinks they probably ignore the tag in that situation.
Q: What’s the fastest way to get millions of pages out of the index? Robots? Redirect? 404?
Maile: Noindex on those pages. Robots won’t do it, and don’t put a robots so it can be crawled and noindexed.
Q: We have a site that ranks for a foreign keyword in both English and Portugese.
Maile: So the same keyword ranks on 2 different pages in different languages. In Brazil, the page in English and the page in Portugese both rank? (He says yes.) She says this is the perfect example of when to use hreflang. Only use canonical when the laugnages are thes ame.
Q: How does Google determine what language content is in?
Maile: Google does that on their own.
Q: What if I have url.com and the page can be served in any of several languages?
Maile: This is not good practice. Separate each language out on different URLs.
Adam’s presentation is 3 Ways to Get Jiggy. It’s about pagination, which is a gnarly problem in SEO. He’ll be reviewing 3 methods: Classic SEO pagination techniques, View All (the best way but hard to get implemented), and the fresh, new method of rel next / prev.
Classic pagination method isn’t recommend anymore because the fresh method works well. Component pages odn’t pass equity and don’t compete in search results. This was fine until HTML 4 and 5 stuff.
Rel pre / next consolidates scores for all pages in a series. No index keeps pages out of index as page 1 is where the focus of content is. In tests its working well. They’re watching how it progresses. It gives the most control. Unfortunately Bing is left out as it can’t follow. You have to use other signals for Bing to accomplish the same things.
Q: Latency issues with View All pageination strategy?
Vanessa says that’s one of the bigger issues. Adam says under 3.5 seconds is ideal. 4+ seconds will show a downhill user experience. The fastest they’ve seen on a big e-commerce was 2 seconds. Maile says that if you can progressively render, that’s going to look good. Vanessa says “perceived” loading time makes a difference.
Jonathan is up next. He says we’ve talked a little about speed. It’s a usability issue. An ad quality factor. Fixing speed issues can be hard to get buy-in and can get deep into site infrastructure. He brings up www.cloudflare.com. This company built a free content delivery network. It grabs static resources, pre-locates them around the net close to where users are. It’s like what akami does for lots of money, but cloudflare does it free. They’ve also built security into the CDN. If someone comes to your site who’s a user, they get your site. If it’s a nasty bot or hijacker, et al, they block it. This speeds up a typical site by 40%. Take a look at it.
Q: Does it hurt your site to use sub-domains? Are you splitting value?
Jonathan: It doesn’t help to change your URLs. Think of the user.
Vanessa: She doesn’t think it makes a difference to use a subfolder or subdomain. You wouldn’t want to cahnge from one to the other just to change it.
Adam: Maybe sub-domains don’t have the same ranking ability a primary domain will have. User experience is the most important thing to think about.
Maile says a blog post about this subject is on the Webmaster Center blog.
Q: How important is it to remove 404 errors in WMT?
Maile: It’s a heads up to you that pages are 404ing. There’s no need to remove them. Links to 404 pages isn’t a great user experience. 404s do not signal lest trustworthy.
Jonathan: Look for 404 errors coming from other sites. Contact the webmaster and have them fix it.
Vanessa: You can focus on large numbers of links to 404s. Don’t worry about every one-off 404 to the site.