Traditional matchmakers who charge a fee only after you decide to get married, have much better aligned incentives.
If they get one month of revenue and you are happy, they made a profit from you for one month and there are no drawbacks.
If they didn’t charge you enough to profit from one month of usage, that’s on them for undercharging.
What fee is profitable in this scenario, for running a short-term subscriptions business? $50/mo? $100/mo?
With any subscription model, satisfying you immediately literally gives them the worst possible financial outcome - matched only by scaring you away immediately.
The worst possible financial outcome is that no one pays anything.
The best possible financial outcome is dependent on your goals.
Do you intend to maximize profit for a 3 month period and then shut down and take your profits to the bank? If so, then you set up the app, sell 3-month subscriptions, and shutdown after 3 months. You will be reviled by all for your tactics, and you'll probably be sued for refunds, but you will have achieved the "best possible financial outcome" for your goals.
Do you intend to maximize revenue over a 10 year period by getting VC investment and going net-negative every month until someone buys you out for your customer database? If so, then you're competing with Facebook, and you will either succeed (because Facebook buys your dataset) or fail (because they don't need to buy your dataset). Your users will suffer the lemon market described upthread because you can't convince the VCs to spend enough money on human verification, and you'll find yourself mocked in the press when your cheapo validation AI rejects 5% of your userbase for having legal names the AI finds offensive.
Do you intend to profit by 15% from every customer for however long they are a customer, and continue profiting in this manner for as long as you have customers? If so, then you set up autoscaling, attract customers, and scale up or down as necessary based on their matchmaking successes.
What, precisely, is your "best possible financial outcome" for a dating site? If it is to maximize revenue growth, then you will fail, because as you pointed out succinctly, it's not possible to maximize revenue in dating without working against yourself.
This could be something that one could actually disrupt. Problem with Tinder is that its interest is not aligned with users who seek long term commitment. Tinder needs its users to stay in the swiping pool to create revenue. The users (at least some of them) on the contrary wish nothing more than finding the right partner and stop using Tinder.
An anecdote from Bumble app, whenever a user finds a partner and therefore stops using the app, the app notices this and starts to aggressively match the user with new partners. This sudden influx of new potential partners makes commitment hard for some users.
If the incentives would be aligned, the user would pay for finding the long term partner and for abandoning the app, instead of paying for super likes etc.