Why reviews beat forum posts as evidence
Three properties, and no other free source has all three at once. The author is a customer, not a browser — they got far enough to install and often to pay. The complaint is attached to a named product, so the supply side of the analysis is already done for you. And the review is dated and ratable, so momentum and severity come with the text rather than having to be inferred.
The fourth property is the one people miss: reviews are written after the purchase decision, not before it. A forum post asking which tool should I use is someone shopping. A one-star review is someone who already chose, already paid, and found out. Only the second kind tells you what the market is actually short of.
Which ratings carry the information
Not the one-star tail, or not only it. One-star reviews are dominated by people who could not log in, were charged twice, or are angry about something that is not the product — real support failures, but rarely markets. Five-star reviews are dominated by people who were asked to leave one.
The information lives in two and three stars, and in a specific sentence shape: a compliment followed by a disqualification. Good app, but it does not handle our case. That sentence is a paying customer describing the exact boundary where the product stops working for their segment — which is the definition of an underserved market, written by the person standing outside it.
Separating a category gap from one product's bug
This is the step that decides whether the reading was worth anything. Collect the disqualifying sentences across every serious app in the category, then sort them by the reason given rather than by the app. A reason that recurs against all of them is a property of the category — nobody serves this, and that is a market. A reason that appears against one app only is that app's defect list.
The distinction is not academic. A defect gets fixed in the next release and takes your company with it; a category gap has survived every release every competitor has shipped, which is the closest thing to proof that it is structural.
Count people, not reviews
One angry customer can write four reviews across two apps in a week, and a review count will read that as four pieces of evidence. It is one person. The number that means anything is distinct authors raising the same reason, because independence is what separates a market from an individual having a bad month.
The same rule kills the most common false positive in this source: a coordinated review campaign, positive or negative. Coordinated reviews cluster in time and share phrasing; independent complaints arrive across months and describe the same problem in thirty different ways. The second pattern is the one worth acting on.
Vietnamese reviews are the least contested source
A global incumbent does not read Vietnamese. Neither does almost any research process aimed at that incumbent's market. So the Vietnamese storefront of a category holds detailed, dated, first-hand complaints from paying customers that essentially nobody on the supply side has read — and a segment nobody has read about is a segment nobody is currently serving better.
This is not a theoretical advantage. App-store reviews are one of the largest sources in this index precisely because of it, and the Vietnamese market's depth here comes from that storefront rather than from forums, which expose RSS but no search.
How this site reads them
The app-store connector answers a market question by asking the Vietnamese storefront which products serve it, then reading their reviews — so a query about a market comes back as first-hand complaints from people who paid the incumbent. Each market hub shows the resulting table: which products people report problems with, counted by distinct people rather than by review.
Individual reviews are deliberately not republished. A single complaint is one person's sentence written in anger, and quoting it verbatim under our own masthead would be republishing abuse at a named company. A count of how many separate people reported the same thing says more and is harder to argue with.