TL;DR: App store reviews are the only large public dataset where people complain about products they already paid for — which makes willingness to pay observable rather than hypothetical. The catch is signal ratio: most negative reviews are about crashes, pricing, and ads. This guide covers which reviews carry information, and which to skip.
Why reviews beat most other idea sources
Most idea sources tell you someone is annoyed. App store reviews tell you someone was annoyed after paying or installing, which is a much stronger statement. The reviewer has already:
- decided the category is worth their time,
- chosen a product in it,
- used it enough to form a specific opinion,
- and cared enough to write it down.
That is four filters applied before you read a single word. Very few public sources apply any of them.
Reviews also have a structural advantage: they are per-product and per-country. You can see the same complaint appear against three competing apps, which distinguishes "this product is bad" from "nobody in this category solves this."
In PainHunt's own dataset, app store reviews across iOS and Google Play are the largest single contributor of scored pain points — over 15,000 in the last 30 days, drawn from 19 countries. That volume is exactly why the filtering discipline below matters.
The four reviews worth reading
Most reviews carry no information. These four types carry almost all of it.
1. The workaround review
"Works fine, but I export everything to a spreadsheet each week to actually get the report I need."
This is the highest-value review type in existence. The user has told you the job, told you the product does not do it, and told you they are willing to spend weekly effort anyway. Effort spent on a workaround is a price already being paid — just in time rather than money.
2. The declined-request review
"Been asking for this for two years. They keep saying it's not on the roadmap."
The gap is not an oversight; the incumbent has decided against it. That is a durable opening, because you are not racing them to build it. Note that the reason is often strategic — it may conflict with their business model, which is the best possible version of this signal.
3. The cross-product review
"Switched from [other app] because of X, but now I miss Y."
The reviewer is doing competitive analysis for you, from the position of someone who has used both. When the same trade-off appears across several reviews of several products, you have found a real axis nobody has resolved.
4. The "wrong customer" review
"Great for individuals, useless for a team of three."
A product succeeding with one segment and visibly failing an adjacent one is the clearest wedge in this list. The market is proven; the incumbent has chosen not to serve a neighbouring segment.
What to ignore
- Crash and bug reports. These describe execution quality, not market gaps. A competitor's bugs are not your opportunity — they will fix them.
- Pure price complaints. "Too expensive" without a stated alternative is background noise in every category.
- Ad complaints on free apps. This is the business model working as designed.
- One-star reviews with no specifics. Emotional intensity is not the same as information.
- Review-bomb clusters. A sudden spike of identical complaints usually follows a pricing change or a public controversy and reflects a moment, not a market.
A repeatable process
- Pick a category, not an app. Individual apps have idiosyncratic problems; categories have structural ones.
- Read the 3-star reviews first. Five-star reviews carry no complaints and one-star reviews carry mostly emotion. Three stars is where "I like this but" lives — the most informative sentence structure in the dataset.
- Read across three competing products. A complaint that appears against all three is a category gap. Against one, it is a product flaw.
- Check more than one country. The same app often fails differently by market, especially around payments, language, and regulation. A complaint that only appears in one country is frequently a local-compliance gap — smaller market, far less competition.
- Count workaround mentions. This is your ranking metric. Not review volume, not star average — how many people describe doing manual work to compensate.
- Confirm the gap is still open. Reviews are timestamped. A complaint that stops appearing after a certain date was probably fixed. Check the recent ones before you build anything.
The honest limitations
Reviews skew negative and emotional. People write reviews when they are angry or delighted, rarely when things are fine. You are reading the tails of a distribution, not its middle.
You cannot quote them. App store reviews are user-generated content with real authorship. Use them to find and count patterns; do not republish the text. (PainHunt aggregates and scores them, then links back to the source rather than reproducing the content — for the same reason.)
Volume is not intensity. A thousand people mildly wishing for dark mode is worth less than twenty people describing a weekly manual workaround.
Selection bias by platform. iOS and Android review populations differ by geography, price sensitivity, and category. A gap that looks universal on one platform sometimes disappears on the other — which is itself worth checking before you commit.
Related reading
- Where to find SaaS ideas in 2026 — the other nine sources, and how they compare
- Pain point research: a practical guide for founders — how to score what you find
- How to validate a startup idea — what to do after you have a candidate
- Search scored complaints across app stores and other platforms in the PainHunt dashboard, or test a specific idea