Recently the folks over at Google had a new algo tweak that once more sought to be faster and more relevant to the users needs. This seems to also be a continuation of last year's need for speed, with elements such as real-time search (now defunct to a large degree), site speed as a ranking factor, and the Caffeine update.
Today we'll look at some of the past geekiness (a.k.a. patents and such) from Google to get a sense of how search engines look at temporal elements. It should be noted we're taking a broad visual, not specific to the recent update as they have stated it's not about updating old content, more about attribution. That's not to say updating content can't be valuable (in SEO) – it's just not part of the recent updates.
Google Patents on Temporal Data
Originally there was a single patent. Then it dissapeared and came back as four patents. Then it surfaced again in the single variety. For reference, here's the ones of interest:
- Document scoring based on document inception date (including our pal Matt Cutts, filed November 2006, published April 2007)
- Document scoring based on document content update (filed November 2006, published April 2007)
- Document scoring based on query analysis (filed November 2006, published April 2007)
- Document scoring based on traffic associated with a document (filed November 2006, published April 2007)
- Document scoring based on link-based criteria (filed November 2006, published April 2007)
FULL (includes all the above); Information retrieval based on historical data (filed December 2003, published March 2008)
That last one is pretty much how it sounds, looking at the amount of updating a page might have. Once more, not (apparently) part of the recent change, but we want to cover all the bases.
Understanding Temporal Factors in Search
Next up, let's get into some of the elements we commonly see in these offerings. We're looking to gain insight into the mind-set of the IR (information retrieval) engineer's head space, not looking for magic bullets and we're not chasing algorithms. We're understanding the concepts of search so that we can better perform and measure this thing of ours.
To get a sense of what they might look at, consider:
- Document inception dates; attribution
- Document content updates/changes; how often a page changes
- Query analysis; brand searches and CTR on queries over time
- Link-based criteria; (see below)
- Anchor text; has page focus changed?
- Traffic; CTR and historical traffic levels
- User behaviour; implicit and explicit feedback
- Domain-related information; registrar and NS changes
- Ranking history
- User maintained/generated data (e.g., bookmarks and/or favorites)
- Unique words, bigrams, and phrases in anchor text
- Linkage of independent peers
- Document topics
There is certainly a vast potential for search engineers when it comes to temporal data. Keep in mind, what we see in isn't always what's going on. We're after the mind-set remember? Take each of the points and weight how well your SEO/content programs are satisfying each.
Historical Data, Freshness and Links
Up next, what the masses drool over: link based criteria. Yes, "freshness" isn't always about the actual content. It can be used to valuate links, too!
Link velocity valuations include:
- When links appear or disappear.
- The rate at which links appear or disappear over time.
- How many links appear or disappear during a given time period.
- Whether there is trend toward appearance of new links versus disappearance of existing links to the document, etc.
Historical factors can also be used in weighting the links such as:
- The date of appearance/change of the link.
- The date of appearance/change of anchor text associated with the link.
- Date of appearance/change of the document containing the link.
- Trust of the document/site where the link resides.
- Freshness of the document.
- Document site authority.
Now, let's consider that the above link related factors are but a layer. This is (yet another) reason why people shouldn't become too fixated with PageRank. There is far more in the world of link valuations. Don't become myopic.
On the "freshness" side of things, they note a page can become "stale" when the following occurs;
- The date at which one or more links to a document disappear.
- The number of links that disappear in a given window of time.
- Some other time-varying decrease in the number of links (or links/updates to the documents containing such links).
This type of valuation is certainly interesting because it most certainly would be a good way of establishing temporal virility of a page. Two important concepts are link velocity and link decay. Essentially the rate they are created and how quickly that dies off.
Looking to get a hold via the QDF, (query deserves freshness)? Play grab and hold via link velocity and decay.
What it Means to You
There is actually a lot more that goes into this area of study. I would certainly spend some time with the above documents and see where it leads.
What is most important to stress is that historical elements aren't just about "freshness." Your SEO (and marketing) programs should always be aware of what we've looked at today as it crosses many lines from link building to content programs. It's no one trick pony.
- Visibility (and indexation)
- Content changes
- Behavioral / engagement
- Domain / registrar changes
- Changes in link focus
- Link velocity and decay
And we haven't even touched on how they might use it for spam.
Many of the noted methods may or may not be in use. Even if we don't know the weighting of those that are, understanding them is essential to modern SEO.
We've seen the need for speed since late '09 and obviously the world of social has made "fresh" the latest in-word. If you're still caught up in the world of the link-obsessed, think again.
And if you missed it, last month we talked about personalization and search. There really is more to SEO than links.
'Til next time - keep it geeky!