Start from a search, not from a feature
The most common way an app fails is not bad code. It is that nobody was looking for it. An idea that starts as "wouldn't it be cool if" has no evidence attached to it, and you will not find out whether anyone wanted it until the app is finished.
Turn the idea into the words someone would type into the App Store search box. Not your product name, and not a feature: the job. "Split a restaurant bill." "Track what I watch." "Learn guitar chords." If you cannot write that phrase, that is itself the finding, because your future customer cannot write it either.
Then check whether anyone types it. A phrase with real search volume is an audience that already knows it has the problem, which is the cheapest kind of customer to reach.
If you cannot write the search someone would type to find your app, they cannot type it either.
Read the competition as products, not as a number
A crowded search is not automatically closed. What matters is who is holding it and how well. Ten polished apps from funded teams is a wall. Ten neglected apps with two-star reviews and a 2019 screenshot is an opening, on the same search with the same volume.
Open the top few results and use them. Read their recent one and two-star reviews, which is where people say exactly what is missing. A repeated complaint across several competitors is the clearest product brief you will ever get, and it is free.
Narrow until you can win something
The instinct is to go broad, because a big market sounds like a big opportunity. In App Store search the opposite is true: a broad term is where the incumbents are strongest, and ranking 80th for a popular search sends you nothing at all.
Narrow the idea until there is a specific phrase you could plausibly rank in the top few for. A recipe app competes with thousands. A recipe app for one-pot meals cooked in under twenty minutes competes with a handful, and the people searching that phrase want exactly that.
You can widen later. Ranking for something narrow and real beats ranking nowhere for something broad.
Ranking 80th for a popular search sends you nothing. Ranking 3rd for a narrow one sends you customers.
Check it can make money before you build it
Look at how the apps already serving this audience charge. If every one of them is free with ads, that is what the audience has been trained to expect, and a paid app walks in at a disadvantage. If several charge a subscription and have healthy review counts, somebody has already proved people will pay.
This is a check on the market, not a forecast of your revenue. What it rules out is the worst case: building for an audience that has never paid for anything.