The Periodic Table of SEO Ranking Factors: 2013 Edition #SMX #11A
Moderator Danny Sullivan, Search Engine Land, @dannysullivan
When they were updating the Periodic Table of SEO Success Factors, they considered adding a “mobile” element, but they didn’t. Google just won’t rank you as well if your mobile experience isn’t good.
Matthew Peters: How Search Engines Determine Relevance
If you search “philadelphia phillies” a search engine is going to decide the results that are relevant. There’s a difference between ranking and relevance. Relevance is the first step – what pages are about this query. Ranking is the second step, when the relevant results are ordered. He’s going to focus on the first piece: deciding relevance.
A search engine computes similarity scores across page elements, title tags, H2, things in bold or italics, body text, all rolled up into an aggregate score. This is a well-studied problem in the computer science and information retrieval field. He’s going to focus on the fundamental ways search and retrieval are done.
Measuring query-document similarity:
- basic building blocks like language identification, spelling correction, word segmentation, tokenization (reviews, reviewer, reviewing > review) + normalization
- higher up the chain are query expansion, classification (images, video, news), local, user intent
- further abstraction are entity extraction and topic model (LDA)
(fancy formulas I missed…)
Which method performs best? The language model performs the best. It’s a new model. When stemming is calculated in, then TF-IDF and language model perform equally well.
- Query-doc similarity is based on decades of research.
- With sophisticated query and document models no need to optimize separately for similar words, e.g. “movie reviews” vs. “movie review.”
- Each page is relevant to many different keywords, so optimize each page for a broad set of related keywords instead of a single keyword.
- Content creation: what keywords will this new post target?
Jenny Halasz: Why Technical SEO Still Works
Crawling issues, signs to look for:
- Consistent crawl patterns
- Increased crawling when changes are made, decreased when site is static
- Big spikes or drops can indicate a problem
3 seconds = good load time
3 seconds = average user attention span
Radware study (Spring 2013):
- Pages are getting slower and larger
- Browsers are not keeping up
- Many site owners are not implementing potential performance improvements
URLs and architecture:
- Don’t create pages just for the sake of keywords
- Clarify which pages have value (XML Sitemaps)
- Keep it simple
- Google is your boss.
- User intent comes first.
- Speed matters.
Marcus Tober: Spam, Penalties and Algorithm Updates
Google has bajillions of pages in its index: 30,000,000,000,000.
Matt Cutts has several methods to fight Internet spam.
Method 1: The red card. Penalty.
Method 2: Algorithm changes like Panda, Penguin, Exact Match Domain – baked into the algo. This is harder to regain rankings from. The number of announced algorithm changes is growing exponentially.
Marcus’s company is doing a ranking correlations study for 2013 that is still in progress. They are looking at number of backlinks, backlinks % with nofollow, backlinks % with stopwords, visibility of backlinking URL…
If you have links from more visible sites you rank better.
From last year to this year, anchor text value has gone down in value
He predicts that if the growth rate stays the same then in February 2016 Google+ will overtake Facebook.
Eric Enge: Social Indexing and Ranking
Who believes Google+ is a ranking factor? Majority
Who believes Facebook is a ranking factor? Not as many
They ran a study on how social sharing impacts SEO. They were trying to find what other things Google and Bing use to discover content. They tested Google+, Google+ shares, Facebook likes, Tweets, Chrome and Google Analytics. They tested on 5 different websites.
Page Performance Tracked Over Time:
- Searchmetrics used for tracking in many cases
- Log files checked to see what pages were accessed
- Backlinks checked as well to make sure people weren’t corrupting data and threw out when they were
- Google+ drives Google indexing and ranking within 4 or 5 days
- Facebook drives Google indexing and ranking within 7 to 8 days
- Links often take longer than this
- Bing did not show indexing and ranking behavior
- Keywords were 3 words or longer in length
- Definitely powerful way to create visibility for content
- Definitely has rankings impact, especially on long-tail and chunky middle terms
- Shows some decay over time
- It’s not powerful enough as an isolated signal, but its a powerful supplement to a traditional SEO strategy