I agree with what you are saying. But we are talking about people who may resubscribe the next month and the month after that, it's not always a one time payment. So if 20 out of the 50 new subscribers play for the next 10 months. That means their subscription did pay for the commercial over time. Not only that. We will have higher #s in the game, thus making the fights more fun and larger, thus bringing in more people.
You are absolutely right. I was just assuming that amortization was rolled into the final ROI analysis.
I'm sure HTC has good stats on the percentage of trials that convert to subscriptions, and the average month player lifespan...accounting for time-value of money, and risk.... The longer it takes the recoup the acquisition costs essentially equates to risk. If it takes 15 months to break-even on the advertising costs to acquire a user, that is more risk than if it takes 6 months. A lot of things could happen, job loss, marriage, school, etc that could cause him to drop out before HTC has broke even on the costs of getting his business.
In the end HTC can calculate a reasonable estimation of the amount life-cycle revenue they can expect on average from X number of downloads. They can then compare that expected revenue to the costs of the advertising needed to generate those downloads. You can't do that for too long without that ratio being a positive number. Unless of course if you are Tesla.
The second argument you made is more interesting. This is a special business because you are not just selling a piece of software, you are selling a massively-multiplayer experience. There is a bit of black magic in the idea of "critical mass". You could argue that you need to burn money for a bit to achieve critical mass so that it becomes self-sustaining. That's scary and un-quantifiable. I'm glad I'm not having to bet my money on black magic.