Want to Rank in Google’s Artificially Intelligent Search? Know the Difference between Proof vs. Relevant Terms
One of FabCom’s higher education SEO & business analytics clients asked why we wanted to update the copy of one their degree offering pages. We explained that while the page contained plenty of keywords that are relevant to popular search topics, it didn’t contain proof terms that indicate authority on those topics. That answer led to another question, “What is a ‘proof’ term and what is a ‘relevant’ term, and why do we need them?”
I thought the answer would also be useful to our readers.
The difference between proof terms and relevant terms
Prior to our update, their Game Programming Degree page covered elements of game programming, but failed to mention the specific names of those elements. The page cited relevant topics, such as “Apple development,” but not the learned skillsets that would normally be part of a discussion about the topic of Apple development. These individual skillsets are examples of proof terms.
Relevant Term: over-arching topics that are commonly searched
and semantically represent a body of subtopics
Proof Term: a subtopic of a relevant term. The presence of
proof terms on a page indicates a very high likelihood that the author knows
what he is talking about and may well be an authority on the topic of which
the term is a subcategory.
The mere presence of matched relevant search terms on a page is no longer solely sufficient for evaluating pagerank, so our team went in and added supporting proof terms to each topic.
How to know which proof terms to add
The topic of Game Programming is a vernacular roll-up term comprised of several proof terms that happen to be programming languages used in developing Apple software, namely:
- Proof term: Objective-C (mostly replaced by Swift in 2014)
- Proof term: Cocoa/Cocoa touch
- Proof term: iOS
- Proof term: Xcode
- Proof term: Swift
- Synonymic proof term: Object-oriented programming
- Synonymic relevant term: game coding
If we look at user-generated sites like Quora and Reddit, entire threads written by a microcosm of experts dedicated to Objective-C make little mention of Apple development because it is implied in the community involved in that esoteric conversation.
Before we included the proof terms, the client’s Game Programming Degree page would never rank for the stand-alone term Apple development degree without branded keywords, because Google had no proof that the page is an authority on the topic. If it were an authority, it would likely be discussing or at least referring to the primary programming language skills that make up Apple development.
How Google knows the difference
The functional technology behind matching a query’s semantic intention to accurate results comes from a patented update to Google’s Hummingbird algorithm that is foundational to Google’s machine learning artificial intelligence apparatus, RankBrain. The update uses a combination of synonym rules that predicts up-and-coming lingo, a floating context database, syntactic context, and search history to assign confidence value to the synonym-paired term.
Synonym Identification Based On Co-occurring Terms (Google Patent
In some implementations, the synonym rules database 185, the adjacent
context database 192, and the floating context database 194 can be included
in a single synonym database for use by the synonym engine 180. In this
case, the synonym rules data, the adjacent context data and the floating
context data can be labeled with markers associated with the data in order
to distinguish each data type in the synonym database…
We will update this post with user signal performance analytics in the second quarter of 2016. In the meantime, you can read some of FabCom’s other case studies and white papers that incorporate SEO and marketing business intelligence.