My perception is that if you don't have access to ~$2M for that kind of gTLD don't even waste your time.
you may want to consult with a practitioner in the jurisdiction of your choice who does business organization and investor equity structures, as the cost to acquire a right to contract for a delegation in the iana root for a string likely to be in a contention set is unbounded. your ~$2m may be a total loss of $185k plus the time value of capital and a "slot" in the 201x evaluation cohort (nomianlly 500 units of evaluation, subject to some grouping if exceeded, for which there are art issues), or it may be a significant over-capitalization with excessive loss of initial investor equity paying for mezzanine or later capital which may be deferred and acquired at lower equity costs after the right to contract has been secured. pat answers are easy. your milage may vary if pat answers are best answers.
(Does the $185K include the given gTLD cocktail party at the seasonal ICANN meeting ?)
this is really a auction avoidance and/or investment invitation cost, and again, ymmv. does gmo improve the likely outome of its campaign to obtain "shop" by holding high cost socials? i don't know. for all i know a bid for .shop will be made by an entity capable of outbidding gmo, or gmo plus its line of auction risk capital and credit, and acquire both "shop" and the value of gmo's pre-auction marketing investment. there's a reason many people use "shoe" as an example of a generic. disclosure only increases risk and cost, except where a rule exist that pre-empts the allocation by auction rule. full disclosure: i spent last fall writing a memo making the case for the affirmative support or non-opposition requirement for iso3166 state capitals to be expanded to all city and geographic names where public administration(s) exists, on secondary meaning (trademark), contrary to the position of the board last fall, adding the public interest by secondary meaning (marks) to the gac's (sovereign) claim. the issue was vigorously discussed at the cartagena meeting, and i think(*) the final applicant's guidebook has the affirmative support or non-opposition requirement for the names of urban areas and other administered regional identifiers. where booze-pub adds value, by all means, spend $$ on booze-pub. where it does not, consider alternatives. -e (*) i've actualy been too busy doing policy work related to applicant support to catch the kurt show in its primary (saturday gnsoc) showing, or any of its reduced calorie repeats during the week, and i've not even opened the .pdf. all that can wait until next week, so for all i know my dagv1/v2/v3/v4/v5 data needs a refresh run over the redline, later.