I have actually seen this graphic before as I was doing research for a class I was asked to give.
From the information that I could gather, this is what I found out. Keep in mind that as the algorithms are updated this is all subject to change as Google keeps their SEO information very close to the vest.
They now look at site analysis, how the site is designed, frameworks, structure, add on, etc. That is compared with the site traffic to determine an internal ranking factor, how they calculate the factor I have not been able to determine. Obviously this is something that they will not release as it could be used to manipulate rankings.
Then those values are compared to the actual site content and the back links. As best I could determine these are used to see if the ranking, traffic and analysis compare to the actual content and the back links. This actually makes sense to see if the ranking are being manipulated. Manipulated rankings will normally not have the traffic or the high quality back links that would be needed to get the the actual ranking. Again, as best I could determine, this is kind of a spot check on the ranking to see if the ranking makes sense based on the content and back links.
Then a determination is made if the keywords pertain to the content of the site. As most of us know manipulation of keywords was a big problems in site rankings for many years and as such they have a lot less impact on SEO then they use to. But again, as best I could determine they look to see if the keywords entered pertain to the content.
Then an SEO factor is established for the site.
What I found most interesting about this graphic was the last few steps. You will notice that it goes through site architecture. I have noticed over the last 18 months or so, that many of the sites that I was developing for customers, were receiving penalties for improper use of tag structure. I had never seen that before.
With some of the tools I was using to check the sites prior to going live, I would see penalties for improper heading structure, alt attribute usage, improper link structure, mobile responsiveness etc. I had never seen these types of penalty points before. The majority of them pertained to tag structure, mobile responsiveness and accessibility.
When I was doing research into this I ran across a couple of articles from the IETF (Internet Engineering Task Force) that talked about accessibility, mobile responsiveness, HTML5 tag and attribute structure, etc. The articles talked about SEO and how HTML5, when fully implemented will impact the SEO of a site. It appears based on some of the penalty points I was seeing that these are coming to age in some the search engines, both Google, Bing, Yahoo and Ask. I have not seen it some of the other engines yet, but they are probably not far behind.
This should be a big concern to anyone using some of the more popular CMS systems, as those systems are designed in a one size fits all architecture. There are 100's sometimes 1000's of lines of code on pages that do not pertain to the actual site. Based on what we are seeing with the HTML5 architecture considerations that would impact the site architecture calculations.
As I am still working on writing the class, this research is far from complete. And as we all know the algorithms for SEO are constantly being updated to circumvent anyone trying to game the system.
Just thought I would share what I had seen since the question came up.