In biz speak (or probably in non-linear programming too), i guess it would be where the different curves meet, the converging point of per capita potential revenues, the volumes (current and future), resources required to tap the market.
Practically this is how the decision tree may look like:
if one has less resources, then it may be wise to tap the market that has the highest per capita of potential revenues(while having reasonably enough volumes)
If one has medium level of resources, then still tap the high per capita market first with the least amount of resouces spent, then once the biz model is validated, deploy the resources for the future biggie market (high vol, low curret potential per capita revenues)
If one has billions to throw like Google, then of course, throw the money in tapping the future biggie and buy off any of the small amart start ups in the developed markets.
You will need to have your European offerings checked by a lawyer, but with the EU harmonization it might not be that big of a deal.
With an Anglophone site, you can reach a substantial portion of the internet.
After that, internationalization and localization takes an awful lot of work and resources, for increasingly smaller piece of the internet. Do you focus on rich, internet savy countries like Holland or Belgium? Or, do you focus on the next two largest languages online: Mandarin and Spanish?
And, then, there's always the problem that the US tends to be a bit enthnocentric. And, companies really don't think about expanding overseas until they are relatively large.
Finally, our educational system is very Anglophone centric, and topics like i18n and l8n aren't really covered. I know they weren't for me, and I finished up 3 years ago. So, the learning curve for a lot of developers here in the US is fairly steep, so they tend to ignore the problem.