I have a colleague who has been involved in two successful H2020 (or one of them might have just been a big EU/EC one) applications (though they have been research grants rather than a prize)

Some tips:
    - the fact you use software developed in other EC project will be a plus btw (he is referring to secrets which has inspired mmt stuff)
    - I won't be in the jury but can well guess the criteria :-P
    - gather proofs, indicators, numbers but also interviews
    - yes you need to build a library
    - i recommend you do your best on the project, applications do not need to be long. they expect short applications with quantitative indicators and a good qualitative narrative
    - the prize is different, not a research grant. you need to demonstrate ongoing success and aim for more
    - write a good application, use the same terms on the bid is always a good guideline.