The biggest differentiator between a borderline fail and a borderline pass is whether the candidate is able to demonstrate that they understand what they're doing.
This means:
Paper 1 - make sure you add in reasonableness checks that show you understand the issue. Before you work out an answer, think of how you'd estimate it, or come up with a back-of-the-envelope calc to get close. Then, compare that to your actual answer. Include what you've done in the audit trail (and in the model).
Paper 2 - conclusions are key. Ensure that you try put down as much as you can about what the results look like, and then ensure that you explain WHY they look like that. Never say 'as expected' without saying WHY you expect them to be that way. Don't say "changing the mortality assumption results in an increase in the reserve required", say "Increasing the mortality assumption results in a higher reserve required. This is because the company would end up paying more out in death claims."
Next steps can help as well - most candidates just bring out the usual suspects (sensitivity runs, peer review, validate info, etc) with a few half-decent ones. To differentiate yourself, show that you understand the problem you're working on by coming up with things that are relevant to that particular problem, or would add something to the analysis. ALWAYS make your next steps specific to the question at hand. Don't say "sensitivity test the key assumptions", say "sensitivity test the expense assumption, as this would have a noticeable impact on the choice between option 1 and option 2". Don't say "Include allowance for tax", say "Include allowance for tax on the investment income earned, and investigate whether tax relief is available on expenses incurred". You get the picture...
The second best thing you can do is ensure that you complete Paper 1 to a reasonable standard before going back to make it better. Getting a great intro, overview, data check section doesn't help you pass if you miss out on half the calculations or half the audit trail. Paper 1 is tight on time - focus on finishing, then polishing.
