I would only recommend doing this on an exam where you can expect to get 90%+ (e.g. GCSE, A-Level, Senior maths challenge if you are strong at it) assuming minimal mistakes. On the aforementioned exams each question is of roughly equal weight, because each question is a question you both should be able to complete and will be able to complete with enough time, so it's sensible to spend roughly T/n time for each question where T is total time and n is the number of questions (or, number of marks allocated to a part of the question as a total of all questions attempted).
The NSAA is apparently difficult enough that applicants get around 50-60%, which means I wouldn't recommend this strategy.
It's hard for me to understand exactly what part of the exam is difficult since I completed my NatSci degree a few years ago so obviously I'm not at the A-level stage anymore. If I had to guess why it's a lot harder, it's probably because most people fall under time pressure and/or the questions are sufficiently difficult or tricky/misleading (personally, I think the questions are certainly
fair even if they are difficult for applicants, there is no useless information, it looks like everything you get given is usually needed to complete the question). The examiners comments on the 2018 paper seem to imply people were running out of time on section 2 (which looks even shorter than 1 given the amount of work needed to do..) so i suspect it is largely time management, which isn't a skill you learn at A-Level (but definitely one you'd learn doing Cambridge exams lol).
The strategy I would recommend would be to first identify which sections you want to do (e.g. A, B, E if physics/maths is your strong point). Do this 100% on what you are best at. Do not think at all stupid things like "i need to do physics because i'm a physical natural sciences applicant" or "I must do biology because i'm a biological applicant" - this is a lot less important than just simply scoring highly. Although I would find section A and E probably the easiest right now, I think A,B,C or A, C, D seem like the most logical choices for anyone who isn't very comfortable with algebra and calculus tricks - I'm sure I would have done A,B,C if I did this test when I applied.
If you are planning on preparing for multiple sections, only bother to decide if you think you are able to identify a section's difficulty in under 2 minutes (and honestly, although this is a skill I developed a lot in 2nd year, I would not have been able to do this at 17). Because the questions are extremely short (usually for MCQs seems to be write down 1-2 equations, solve, get the answer), judging how hard it is usually just means almost answering it anyway. For longer questions in section 2 this may be worth doing though.
In terms of actually answering questions, I would power through the questions as fast as possible. Skip anything which you can't immediately do. This probably means doing around 40-70% of the questions. The questions you can do should take you under 1 minute (since you can do them and they are all 2-5 line solutions at most). Even assuming errors this should bring your score up to 30-50%, and probably in around 30 minutes. The remaining questions can then get your focus for 40 minutes, which probably means 2 minutes per question (approx), easily enough time to solve if you can solve, or to at least eliminate obviously wrong (wrong order of magnitude, etc.) answers and have a guess that is 1 in 2 or 1 in 3. The remaining 10 minutes should be spent on educated guesses to whatever remaining questions, or checking answers to questions you did at the start (consolidation of easy marks). 10 minutes should be enough to check answers to ~25 questions. Doing this would mean a bit more practice and a bit of a different style, but I think it would be very difficult to do badly (under 40%).
Another good option could be to treat Paper 1 as 3 x 25 minute exams with 5 minutes bonus time for whichever part needs it. I think this would work very well if done properly also. The data from the FOI request
https://www.whatdotheyknow.com/request/2017_natural_sciences_offers_and?unfold=1 suggests to me that people heavily over-spent time on section 1A, and were not spending enough time on the optional parts.
The worst possible semi-logical approach (ignoring stupid things like doing questions from more than 3 different sections) would be to do questions one-by-one and not move on if you are stuck. I think this could easily cost you 20%+ in terms of lost time on later questions and lost expected-marks making educated guesses rather than solving questions.
In terms of results, I think the exam is actually quite important (I said otherwise before seeing the data; but the correlation between overall score and offer "success" looks very high. Almost every applicant with an average score above 7.0 (hard to say what exactly this is, but it's probably around 40/54) got an offer. And less than 10% of the bottom ~500 scores (below 3.0 - probably below 17/54ish) got offers. I would have to do proper analysis to deduce anything further but a sensible guess for the coefficient of correlation here may be around 0.4 - which would mean that the exam may be more important than the interview.