How to Find Indie App Ideas by Mining App Store Reviews (Free Method)
Most "find a startup idea" advice is bad.
It tells you to "talk to users" before you have any users. Or to "scratch your own itch" when you don't have an itch. Or to read trend reports and bet on whatever is hot this month.
Here is the method I keep coming back to. Read app reviews. Specifically, the 1-star to 4-star reviews of paid apps in a category I might want to build for.
This is not new. It's not clever. It's not what you'll see in a Twitter thread with 10k likes. But it works, and almost nobody actually does it because it's slow and boring.
Let me show you what I do, and why I think this is the most honest way to find an idea worth building.
Why paid apps, not free ones
Free apps get reviews from users who paid nothing. A lot of those reviews are noise. People who downloaded by accident, complained about ads, or wanted a totally different app. The signal to noise ratio is terrible.
Paid apps are different. The user gave the developer money. Then they took the time to write a review. That is a person who wanted the app to work, used it, and felt strongly enough to type something out on their phone. The thing they are complaining about is the thing that actually matters to them.
This is the simplest filter you can apply to get higher quality demand signals than any keyword tool can give you. It's also free.
Why 1 to 4 stars, not 5
5-star reviews mostly say things like "Great app!" or "Love it" or "Best in this category." They are pleasant to read but tell you nothing about what is missing.
1-star reviews tell you what is broken. People are angry. They are specific. "Crashes when I try to import a CSV." "The sync stopped working after the last update." "I bought this for the calendar feature and it doesn't even support recurring events."
But the most useful range is actually 3-star and 4-star. These are the reviews from users who like the app enough to keep using it, but want something that isn't there. "I love this app, but I wish it had X." That is a feature request from a paying customer who is already invested. If you can find a pattern of the same X across many apps in a category, you have found a real product opportunity.
The actual workflow
Here is what I do. It takes a few hours per category and you can do it on your phone or your laptop.
Step 1. Pick a category. Open the App Store. Go to the Apps tab. Pick a category you have some context for. Productivity, Health & Fitness, Photo & Video, Education. Any of these have a long tail of paid apps with engaged users. Avoid Games unless you specifically want to build a game.
Step 2. Find paid apps with reviews but not too many reviews. This is the important part. The headline apps in any category, like Notion or Things or Day One, are saturated. Their pain points have been studied to death. You want the second tier. Paid apps with maybe 100 to 5,000 reviews. Enough that there is a real user base, but few enough that no big company is paying attention to the complaints.
You can sort by category in the App Store but the sorting is not great. Honestly the easiest way I've found is to search for category-related keywords and skim the paid results. Or use a third-party app discovery site if you have one you like.
Step 3. Read reviews. Open the app's page. Scroll down to reviews. Sort or filter to non-5-star if the option is there. On iOS this is hidden, so you basically have to scroll past the 5-star ones. Read the 3 and 4 star ones first, then the 1 and 2 star ones.
You are not reading to evaluate the app. You are reading to find a phrase that repeats. Maybe three different users say "I wish there was a widget." Maybe two of them say "the export is broken." Maybe four of them say "no offline mode." Write these down. Just a list, no formatting needed.
Step 4. Repeat across 5 to 10 apps in the same category. This is where the magic happens. One app having a missing feature is not a signal. Maybe that one developer just doesn't care. But if you check 8 paid task management apps and 5 of them have users complaining about the same missing integration, you have found something real.
Step 5. Look for the gap. Now you have a list like this:
- 5 of 8 apps: users want better calendar sync
- 4 of 8 apps: users want offline mode
- 3 of 8 apps: users complain about the export feature
The thing that nobody is doing well across the category is your candidate. Not a guaranteed winner. You still have to check if it's possible to build, if people will pay for a separate app for it, all that. But you have a real signal that the demand exists, from real users who already paid for something in this space.
A small example
I'll show you what this looks like in practice. I picked a random category, Health & Fitness, and pulled up a few mid-sized paid apps. I'm not going to name them because the point is not to dunk on these specific apps. But the patterns I saw, after about 90 minutes of reading:
Across 6 paid fitness tracking apps in the $4 to $10 range, with review counts from 200 to 3,000:
The most common 3 and 4 star complaint, by far, was about Apple Watch sync. Either it didn't work, or it worked but was slow, or it required some annoying re-setup every few days. This came up in 4 of the 6 apps.
The second pattern: people wanted to manually edit a workout after the fact (fix duration, change exercise type) and the apps either didn't allow it or made it really clunky. This came up in 5 of the 6 apps.
The third pattern: data export. Users who track for years want to get their data out into CSV or some other format. Half the apps had no export at all and users were upset about it.
Now, is "build a fitness app with great Watch sync, easy retroactive editing, and clean CSV export" a guaranteed winner? No. But it's a very different starting point than "I want to build a fitness app." It comes with three concrete features that real paying users have already said they want and aren't getting.
Doing this once by hand is the cheapest way to learn what real demand signals look like. After that, you can either keep doing it manually or look at automated pain point analysis of paid App Store apps to scale it up. The reading skill is the same either way.
What goes wrong
A few honest things.
It's slow. Reading reviews is boring and your eyes glaze over. After about an hour you start missing patterns you would have caught earlier. Take breaks.
The signal is in the boring complaints. The funny rage reviews ("worst app ever, refund!!") are useless. The 3-star "I really like this but..." reviews are the gold. They are also the hardest to keep paying attention to.
You'll be tempted to build for yourself. When you read 50 reviews you start to think "I would never have that problem because I would just X." That's fine, but it doesn't matter. You are not the customer. Trust the pattern, not your own preference.
One category is not enough. If you check one category and don't find a clear gap, that does not mean the method failed. Try a different category. Different categories have very different opportunity densities. Some are over-saturated, some have huge gaps.
This finds features, not businesses. The method is great for finding what is missing. Whether the missing thing is worth a whole separate app, or whether it should be a feature inside an existing app, is a separate question you have to answer.
The obvious automation
Reading hundreds of reviews by hand is the slow version of what tools can do faster. This is what we built AppPainPong for. It pulls 1-4 star reviews of paid apps and clusters them automatically, so you can see what users want but aren't getting without spending a whole weekend reading. The free tier covers the manual work I described above.
But honestly, even if you never use a tool, doing this exercise once by hand for a category you care about will teach you more than any market research report. Reading actual user words in their actual angry or disappointed voices is how you get a real feel for what people want. No dashboard replaces that.
Where to start
Pick one category right now. Spend 30 minutes reading reviews of 3 paid apps in it. You won't have enough data to draw a conclusion from 3 apps, but you will quickly see whether reading reviews actually feels useful to you. If it does, schedule another 2 hours for the rest of the workflow.
The hardest part of this method is not the method. It's sitting down and doing it instead of looking for a faster shortcut that doesn't exist. Once you've done it once by hand, then scaling up with automated App Store review analysis makes sense — but only after.