ReviewFold
Example study · Apple App Store · United States
Threads App Review Analysis: A Research Brief
A saved sample of 200 written reviews, with selected excerpts checked and organised into three editorial research themes. This is not a verdict on Threads. It is a demonstration of what a review sample can support, where it contradicts itself, and which questions are still open once the reading is finished.
Sample · 200 written reviews
Collected September 23, 2026
- 1★73
- 2★11
- 3★5
- 4★8
- 5★103largest group
01 · The rules of the reading
A theme is a group of accounts, not a count of users.
Three habits keep this brief honest. They are also the reason it is short.
-
Every theme names its reviews
A theme is a set of excerpts you can go and read. Two accounts can describe a task clearly; they can never size it.
-
Reviews describe an experience, not the build
These were written between September 13 and September 22, 2026. Anything fixed since then would not show up here, and anything broken since then would not either.
-
Counter-evidence sits beside the claim
Where a four- or five-star review describes the same problem, it is quoted inside the same theme rather than moved to a footnotes section.
The brief is useful for one decision: which tasks to go and check. It is not a scorecard, and it does not rank fixes. Choosing an order is a separate job, covered in which complaints to investigate first.
02 · Theme one
Reviewers want a feed that better reflects their interests.
Three reviewers want different recommendations. One describes trying “Dear Algo”; the other two describe a mismatch with their interests without establishing which controls they used.
-
“Then my algorithm acted up out of nowhere. And now my feed is full of depressing people.” … “And the Dear Algo thing didn’t even help.”
The review names a specific control and says it failed. Whether that control was used as designed is not something the review can tell us.
-
“…the suggested feed just trying to be rage bait instead of being an actual feed that mirrors who I follow and engage with…”
The expectation is stated precisely: a feed that mirrors follows and engagement. This links dissatisfaction with the suggested content to an expectation about follows and engagement.
-
“…it could be better like the algorithm could respond better to what I want to see…”
Four stars, same request, softer wording. A theme that only quotes one-star reviews would have hidden this one.
Still openWhat “Dear Algo” actually does today, and whether the reviewers who mention it used it as intended. The sample records what people believe happened, not what the system did.
03 · Theme two
Moderation feedback leaves questions about explanations and appeals.
Four reviewers describe restrictions or flagged posts. Their concerns differ: warning, explanation, access to help, and room to add context during an appeal.
-
“My Threads account was wrongly disabled by AI with no recourse or human to talk to.”
Two separate claims in one sentence: an automated decision, and the absence of a person to review it. Both remain the reviewer’s account; neither identifies the actual decision process.
-
“…they didn’t warn me or anything they just banded me like I wasn’t real…”
A different concern about warning and account history: the account was verified and had been in use for months. Tenure and verification did not predict the outcome.
-
“They keep flagging posts that fit the guidelines.” … “when you appeal you aren’t allowed to explain what your post it extra sucks”
Here the appeal exists, and the reviewer describes its limit: there is nowhere to add context. The excerpt describes a limitation of an appeal, not the absence of one.
-
“I was suspended and banned permanently for breaking community guidelines, with no good explanation.” … “It made me submit a picture, which was a live selfie and prove I was human.”
This selected account describes a first ban, a second account, a live selfie check, and a second ban after complying. It reads as a sequence the reviewer could not exit.
Still openWhich system acted in each case, whether any of the bans were correct, and what the appeal screen actually offered. A public review records the reviewer’s understanding of a decision, not the decision.
04 · Theme three
Reading and posting are interrupted in different ways.
These are three distinct tasks to check: retaining a reading position, saving a photo-post draft, and attaching a GIF. They do not establish a shared failure or cause.
-
“…switched between apps, the page refreshed and I always lose the content…”
The reviewer is not asking for a feature so much as for the app to remember where they were. They wrote it as a request, politely, at four stars.
-
“never works when posting a full 20 pictures, quits without saving draft”
The reviewer reports a quit without a saved draft. Whether any recovery was available is not established by the review.
-
“I haven’t been able to post a gif saved from my camera roll in a post. The support is AI and doesn’t help.”
A specific file type fails, and the route out is also automated. The reviewer also reports unhelpful support; the cause of the attachment failure remains unknown.
Still openWhether these are common or rare. Two hundred recent reviews are not a rate, and the sample stopped at a page limit rather than at the end of the store’s history.
Counterweight
The same wish, from the other side.
Theme one quotes people who feel the feed ignored them. This five-star review describes a more positive experience of the conversation and content.
“Looking for discourse without trolls and disturbing imagery, look no further”5★ · Sep 17, 2026 · same sample
Read the two together and the theme gets more useful, not less. The positive account does not disprove the complaints or demonstrate that feed controls worked. It shows why the selected complaints should not stand in for the whole sample.
103 of the 200 ratings are five stars. A brief that quotes only the angry reviews is a brief about the angry reviews.
05 · The edge of the evidence
What this sample cannot settle.
Stating the limits first is cheaper than being corrected later.
-
It cannot rate the app
The collector stopped at its page limit after 200 written reviews. This does not establish the total number of written reviews or how representative this sample is.
-
It cannot date a change
The reviews span September 13–22, 2026. A fix or a regression after that window is invisible.
-
It cannot separate app from device
One reviewer writes only that the app “does not work properly on iPads”. That could be layout, a missing tablet build, or one broken screen. The sentence does not say, so this brief does not guess.
-
It cannot name the system behind a decision
Reviewers call it AI, bots, or moderation. Those words describe what the experience felt like, not which pipeline acted.
Sample recorded September 23, 2026 from the public US App Store listing, using ReviewFold’s Apple public-web collector at its two-page limit. Reviews are reproduced as written, including their spelling. ReviewFold is not affiliated with Threads or Meta. Threads is a product of Meta Platforms, Inc., used here only as a public example.
06 · Repeat it on your own app
Three moves, and the third is the one people skip.
- 01
Take the sample you can name
One app, one storefront, a stated count and a date. If you cannot describe how the reviews were selected, the themes will inherit that vagueness.
- 02
Group by the task, not the adjective
“Confusing” is not a theme. “Could not open a post’s menu without opening the post” is. The second one tells you what to check.
- 03
Write down what contradicts the theme
Keep the four-star version of the same complaint, and the five-star review describing a different experience. Without them you have collected supporters, not evidence.
ReviewFold groups related feedback from a recent public-review sample of one app and one region, and keeps the original excerpts attached so you can overrule it. The grouping is a starting point for reading, not a conclusion.
Current support, stated plainly
- Apple App Store: US and Canada. Google Play: US.
- Up to 200 recent public reviews per report.
- One app and one region per report; no cross-app comparison.
- A free preview shows themes and counts; the paid report adds review excerpts, PDF and CSV.
- No sentiment score, opportunity score or market sizing.
Same method, your reviews
Find the tasks your own reviews keep describing.
Paste a store link, get grouped feedback with the original reviews attached. Use the brief above as the standard for what the output should let you check.