Anywhere where you exist on top of someone else's service is basically killed by a 30% margin. To use an example I worked on a while back, dynamic Minecraft server hosting: Taking advantage of the fact that most small servers need much less than 100% uptime, you realize that if you dynamically spin up and down these servers, you can offer substantially better use-based pricing. You're basically an AWS reseller in this scenario, so there's a hard floor in what you how much you can charge, because VPSs cost money, but you're competing against "leave the server on 24/7", which is your upper cap. Unfortunately, unlike those "Up 24/7" servers, you need a good way for your customers to spin the server up and down. A phone app seems like a good way to do that, until you realize that 30% of that revenue potential would be forked over to the app stores.
100% AWS uptime, at cost -> your competitor's price
10-15% AWS uptime -> your marginal costs
50-60% AWS uptime -> maximum price you can really charge while differentiating yourself on price
25-32% AWS uptime -> subtract your costs and google's cut from that maximum price
When you start factoring costs like customer acquisition, it starts getting very, very tight. 30% profit can make for a pretty healthy business, especially in businesses where your prices and costs can be predictably coupled. But a business with a 30% profit margin that has to pay 30% is a failed business.