Yes that certainly sounds a good strategy! Although don't quite understand what you mean by keyword familiarity - do you mean brand names or words specific to a particular niche? Sounds interesting.
To be honest from my experience, I don't really have much trust in any exact method of picking the right keywords initially anymore. I've wasted so much money on software/subscriptions which predicts high traffic/low competition keywords and they've never really worked properly. The only one that seemed to work well was a program a guy from the warrior forum produced (sparrow think his name was) but it stopped working a couple of years ago!. SEMrush keyword difficulty score was about the next best I found.
Now I usually just try and find low competition ones and use them as a starting point. After a while I check which keywords are getting me clicks in GWT and then optimize each page based on those - sometimes they are the ones I targeted, often completely different. It's only at this point that I use the content analyser to fine tune based on those keywords that give me traffic.
Indeed Google has been changing the meta description for many years - I remember lots of them used to come automatically from ODP data from DMOZ directory. Especially annoying when you'd had wait years to get included in the damn thing in the first place. Used to get very frustrating but must admit have always thought it's something I can't control so I just tend to hope mine are used !
The content analyser I mentioned has a lot on user sentiment (which I'm presuming is similar to BERT user intent but could be wrong) - you can track the user sentiment e.g. positive, neutral or negative content against a specific keyword. You can even track the correlation between the sentiment and position in rankings. I can't quite figure out to do with this information or even whether it's really valid. You also have to use extra credits to track this so I don't always bother.
For example I had one keyword where my page was reported as having a strong negative sentiment for a keyword and ranked at number 7. All the other pages in top ten had a strong positive sentiment - the suggestion from the data was that I should change my page to be more positive. I did a test by trying to change the NLP sentiment by using this site - text2data.com. I really didn't know what I was doing but swapped out negative words for positive ones based on this. My site initially dropped lilke a stone from 7 to about 67 for that keyword !! However after 8 weeks it now seems to have settled at position 3 for a reasonably competitive keyword and is getting a lot more traffic.
I'm not sure I'm convinced yet (as my test has only included one site so far!) and definitely still don't really understand it though.

However the idea of significantly boosting a page's traffic with very little effort and no expense certainly appeals ! So will probably try again !