The rest of the on-page factors are far less important than the title tag, in terms of ranking, so much so, in fact, that I relegate them to mid-tail or long-tail influencers only. For these far less competitive searches, the search phrase alone is sufficient to garner decent ranking. Coupled with a couple of decent inbound links and top ranks are often trivial.
In our meta description, we also get far more words and characters than will fit in title tags, so we can target more phrases and make our text read better to humans, a matter especially important when we consider click-through.
The strength of match employed for descriptions and body copy works identically to the matching done on titles, with one significant exception that pretty much changes everything.
In the case of titles, the length limit plays almost as an advantage in that we really only have one place for the keyphrase to appear – not so with meta descriptions and body copy; there’s so much room in both of these that one or more of the keywords or the entire keyphrase could appear multiple times within each of them: which one should be used in assessing the relative ranking of two pages then?
Search Engines – Ya Gotta Laugh
What the engines act like – I’m drawing no conclusions about how this is actually implemented – is that they use a hybrid of two well-known text analysis techniques. I presume they are optimising for speed since this matching is being done at query time while the search is watching, like, the hour glass. Best fit is expensive and requires that the whole page is searched for the keyphrase and then, conceptually (because there are short cuts), read again looking for various forms of partial match or scattered keywords.
Each of these passes is then scored separately and the best result for each page selected.
First Fit is a far cheaper strategy that reads through the text one time and with each new word computes a score and terminates when some good enough condition is met.
Hybrid approaches here are as numerous as there are programmers, but for Google it looks like they’re using the text indexed to jump into the middle of the text where two or more of those keywords in a phrase are clustered together, and then score exact matches in that neighbourhood; this becomes semi-first fit because they tend to grab the first cluster of keywords they find, not the most complete cluster.
This is why the first match of keywords on the page appear to get selected far more often than the latter mentions of the same words, even if the latter mention of the phrase might be a better match.
Some On-Page LinkLuv
On-Page SEO: Hello World!
On-Page SEO: the Title tag
On-Page SEO – Keyphrase Positioning – Display Order v. Markup Order
On-Page SEO – Head and Body – A Foot in Both Camps
On-Page SEO – Head and Body – Both Feet in Both Camps
Related: On-Page SEO – How Many Characters in a Page Title for Google SEO?