My personal suggestions/priorities:
-Multi-level output, perhaps a twist switch for high and a programmable low like the Nitecore or Jetbeam M models. If you are after dive usability, you'll need to come up with a non-twist mechanism for setting the level.
-Smooth beam. Liteflux LF5XT or Nitecore D10 Golden Dragon Plus version (not the EX10, though) are good examples of this. The beam is not devoid of a hotspot, however it has a nice transition. If you want to allow for photographic use, sell it with a Surefire style over the bezel diffuser.
-Batteries: stay away from exotic battery technology (such as CR123A) if you intend to sell in the fairly large non-US market. Too much $$$ to run and many headaches to export and sell. AA torches will have a larger market share than CR123A. 18650 does not have good market saturation yet, despite being a nice capacity and size. If you sell an 18650 torch and want to have it sell well (50K units/year) you will need to have it be a package deal including charger and batteries. This will result in needing certifications in most countries of sale, reducing profit and market share. Fenix TK20 is a good example torch for what can be done with 2xAA.
Non-exotic batteries will also be significantly easier for owners of your product to transport, specifically when it comes to air travel.
-Rechargeable compatibility: Make sure that you account for the differing voltage range of NiMH over Alkaline, or Li-ion over Lithium Primariy if you design a light that supports user replaceable batteries. This has been a reason for why Surefire lights do not well well outside of the USA, where the primary batteries are not as easily attained.
If you manufacture a light that has a built in, non-user replaceable rechargeable battery, keep in mind that the charger for the battery will likely needs its own certification in most countries of sale.
-Build material: Obviously intended use with dictate construction materials. If you want to sell it for optional use as a dive light, you'll need to be a polymer or non-reactive metal construction. The polymer may have heat dissipation issues if used out of the water, so titanium would likely be a better choice in that case. I'm not certain if you can build an aluminium torch with a replaceable reactive anode or not, but it may be worth investigating. Also, titanium would not be neutrally buoyant.
-Durability: SS bezel and tail cap rings are a nice design feature that help protect less hard metals from drops, and they do not cost that much to implement. A good sample implementation is the Jetbeam Jet-III M.
Keep in mind that a crenulated bezel can attract too much negative attention while not necessarily adding much functionality. A slight crenulation can allow you to see if the light has been left on when head standing, without drawing negative attention to the light at an airport.
-Throw: How much throw are you after? Optics have their place, however they dramatically limit usability. Larger reflectors impact on design considerations. Generally speaking, focusing heads introduce more problems than they solve.
-Price point: I'd pay up to about $200 USD for a TK20 sized aluminium high CRI torch with the above features, or about $300 USD if made from Titanium. However, I suspect that 50K consumers per year might not spend that much. $50-100 USD might be a more realistic target price, if this can be achieved. If the light has a special purpose target audience, pricing will vary with the expense of the activity in question (ie, caving or diving.
If you have not already done so and happen to speak German, I'd recommend that you also post over on messerforum.net as they have a decent number of high-end torch owners who I know would have some good input as well.
Best of luck!