“Strings not things.”
“Entities, not keywords.”
“Keywords are dead!”
Sound familiar? It should. Ever since Hummingbird replaced Google’s core algorithm, the Knowledge Graph started showing up in search results, and we kissed our Google Analytics keyword data goodbye, SEOs have been evangelizing about the death of keywords and the birth of semantic search.
And what actionable advice have we gotten in return?
“Pretend Google is a person.” “Optimize for users, not search engines.” “Rich snippets.” “Google+ all the things.”
I don’t disagree with this advice. You should act like Google is watching your every move, and you should put users before search engines. Yes, rich snippets help your CTR, and yes, Google+ is good for getting your face in the SERPs and connecting with people.
But I feel like I’m missing something, because this advice doesn’t seem very practical when it comes to optimizing for this new era of semantic search.
So let’s talk about what search engine optimization looks like in the era of semantic search.
Google: Thinking in Triples
The framework of the semantic web is the triple. Aaron Bradley has an excellent in depth post on triples and their relationship with the semantic web, and I don’t want to spend a ton of time on this, but it’s important that you understand how Google is thinking about “entities” on the web.
A triple follows the basic formula:
subject > predicate > object
Schema.org is built on triples, like this:
review > has rating > 4.5
And Google+ understands things about people in triples, like this:
Carter Bowles > works at > Northcutt.com
And Google even understands things about your site based on traditional HTML in triples, like this:
URL > has title > title of page
Why is this important?
It’s not enough to say that Google is using “semantic technologies” to understand “things not strings.” If you want to optimize your pages for Google in this new era of SEO, you need to understand (at least on some level) how Google is actually interpreting the information it scrapes from the web.
You could just count on Google to do all the heavy lifting for you, but this is putting a lot of faith in the algorithm.
Remember, along with Microsoft and Yahoo!, Google created schema.org because it helps “search engines understand the information on webpages and provide richer results.” Don’t think that schema.org markup only matters when it helps you get a rich snippet in the search results. It’s an important part of how Google interprets what it sees on your site.
If you want more relevant traffic, you want Google to understand your site as well as possible. This means all of that schema.org markup matters, not just the markup that will give you a nice pretty snippet in the SERPs.
This goes a lot deeper than markup, though.
As all the evangelizing has made clear, SEO is now about “entities, not keywords.” But how can we actually use that information?
For starters, we can recognize that it’s actually not just about entities. Let’s take this query as an example:
“Where is the nearest burger joint?”
This search query is not an “entity.” “Burger joint” is an entity. But the user intent is a triple:
user > finds > “nearest” “burger joint”
So, part of the goal of semantic search is to understand that “burger joint,” “restaurant with burgers,” and “burger place” all refer to the same entity. But it’s also to understand verbs like “find,” and even insert that verb into the search query when the user doesn’t explicitly use it. It’s also to understand adverbs like “nearest” or adjectives that give additional information about the entity in question.
Consider the following two queries:
“closest burger place”
“best nearby burger place”
The first one has the same meaning as “Where is the nearest burger joint?” The second one, on the other hand, means something different, even though both are referring to the same entity. It’s asking for the best nearby burger place:
user > finds > “best” “nearby” burger place
This is asking Google to solve a very different problem. Now it doesn’t just have to find the closest burger place. It has to weigh the quality of the restaurant against its distance from the user: and that’s a very different kind of problem to solve.
So, how does this affect keyword research and implementation?
The most important thing to understand is that Google no longer thinks of search queries as nouns.
Google has played thesaurus before and swapped out individual keywords for others that mean the same thing. That’s not really anything new.
What’s new is that Google is attempting to interpret search queries as statements or questions.
These days, you don’t necessarily want to target a keyword phrase like “where is the nearest burger joint?” at all (unless, say, you’re a business directory), or even a synonym of it. This is because Google is now attempting to interpret a query like this as an instruction. Instead, you simply want Google to understand that you are a “burger joint” and that you have a specific location, a specific reputation, and so on.
The more information you can provide Google with, both directly through markup, and indirectly through content, internal links, user behavior, and external links, the more effectively Google can connect you with users who are searching for you. This is where a lot of long tail traffic is coming from these days.
Site Structure: Mirroring the Knowledge Graph
So, the obvious advice here is to target groups of queries that reflect distinguishable user intent. Two queries that mean the same thing should be treated like a single keyword, and you should only create a single page for both of those queries.
The less obvious piece of advice, which we just discussed, is that you need to be able to distinguish between the parts of a query that are meant for Google, and the parts of a query that are meant for your site (and the entities its attached to).
The next question is what your site hierarchy should look like.
In modern SEO, hyperlinks need to be thought of as more than just ways to pass “link juice” around. Hyperlinks help establish relationships between entities. They are actually one of the oldest triples Google has access to:
page 1 > links to > page 2
But links can be more meaningful than this:
page 1 > belongs to > category 1
page 1 > is a prerequisite for > page 2
page 1 > is related to > page 2
I might even go so far as to say that if there is a verb in or near the anchor text of a link, it might establish a more meaningful connection between two pages and their associated entities:
product 1 > “manufactured by” > company 1
Even if Google isn’t necessarily there yet, this is the kind of thing you should be thinking about when you build your internal and outbound links.
Putting Semantic SEO to Use
The fact that Google has embraced the semantic web doesn’t mean that Google will suddenly reward sites for having great branding or marketing campaigns. Without smart implementation of markup, intelligent keyword research, and strategic site hierarchy, you’re putting too much of the burden on Google’s algorithm.