ReviewFold
Guide · Worksheet
App review analysis template
Use this worksheet to keep what a reviewer said separate from the meaning you assign to it. Download the blank CSV, open it in your spreadsheet, and start with one app, one store and one region. No account is required for the blank template.
The file
- App
- Store
- Region
- Collection date
- Review date
- Rating
- Source reference
- Short excerpt
- Task or context
- Statement type
- Candidate theme
- Evidence relationship
- Next check
The order of columns is fixed so that two files from the same team can be compared without rearranging them. The blank file holds no reviews, no formulas and no personal data.
01 · The record
Start with the source.
Record the app, store, region and collection date before you start grouping reviews.
Keep the review date as a separate field: a review written in August and collected in October describes an August account of the experience. If you merge those two dates, you lose the ability to tell whether a complaint is about the product you are looking at now.
Use one row per review. If it describes two different issues, record both in the notes instead of duplicating the row and accidentally counting the same review twice. Keep a stable reference to the original in your own working file — a link, an identifier in your saved sample, whatever lets you get back to the sentence you quoted.
One row, one review
Duplicating a row to hold a second issue inflates every count you build on top of the file. If a review covers two matters, one row with both noted beats two rows that look like two people.
02 · The columns
What each field is for.
Thirteen columns, and none of them asks for a score. The file records what exists; the judgment stays visible as a separate column you can argue with.
- App, store, region
- The listing and market you actually read.
- Collection date
- When you saved the sample.
- Review date, rating
- Source metadata; the rating does not decide the category.
- Source reference
- A review link or a reference you can trace in your saved sample.
- Short excerpt
- The smallest passage that preserves the relevant meaning.
- Task or context
- What the person describes doing. Use “unclear” when it is missing.
- Statement type
- Request, reported experience, praise, question, or mixed.
- Candidate theme
- Your working label, kept separate from the excerpt.
- Evidence relationship
- Supports, challenges, adjacent, or unclear — relative to that label.
- Next check
- The question that would resolve an uncertainty.
A rating and a theme answer different questions: the same review can be five stars and still ask for something the app does not do. Sentiment versus theme analysis reads one review both ways and shows where the two columns disagree.
A blank cell is preferable to a confident guess. “Unclear” is also useful data: it tells you what the review cannot answer.
03 · Filled rows
Three filled examples.
These are selected excerpts from a saved sample of 200 Remini reviews from the US App Store, collected October 21, 2026.
“make it totally free with just adds”
Working interpretationWants an ad-supported free option. The excerpt does not establish that an earlier option was removed.
Next checkWhich task does the reviewer want to complete, under which access conditions?
“Must pay now, no freebies”
Working interpretationReports losing free access; the excerpt gives no detail about an ad-supported path.
Next checkWhat screen or feature was blocked?
“watch an ad, or pay for the subscription”
Working interpretationChallenges a blanket claim that the ad option was unavailable to everyone in the sample.
Next checkWhat conditions could explain the different accounts?
The classifications are editorial judgments, not a claim about the current product or the frequency of these experiences. ReviewFold is not affiliated with Remini. The spelling “adds” is preserved from the review.
These rows are examples of filling the worksheet, not three independent estimates of demand. Read together, the second and third rows describe different access experiences. They do not establish that the reviewers encountered the same feature or conditions — that uncertainty belongs in the “evidence relationship” column.
04 · Before counting
Review the label before counting the rows.
Filter the candidate-theme column and read the excerpts together.
A broad label such as “paid access” may collect different experiences. A narrower label such as “ad-supported access removed” needs more specific evidence. When a label survives that test, it is describing something a reader could check, not a feeling the writer had.
Check the “challenges” and “adjacent” rows before describing the group. Keep unresolved accounts visible. If you report a count, say exactly what you counted: distinct reviews in this saved sample supporting the stated interpretation. It is not a percentage of all users.
- SupportsThe account is consistent with the label.
- ChallengesThe account conflicts with it, or with another row.
- AdjacentRelated, but about a different step or condition.
- UnclearThe excerpt does not settle it either way.
No severity score
The worksheet has no automatic severity score. For deciding what to investigate first, use the separate feedback prioritization guide.
05 · Keeping it honest
Keep the working file useful.
A worksheet that cannot be traced back to the sentence it quotes is a list of opinions with dates attached.
Save a copy of the untouched source alongside your worksheet. Record changes to labels in a note, and use the same meaning for a label throughout a pass. If two people review the same material, discuss disagreements rather than silently combining incompatible categories.
The CSV is a blank research worksheet. It does not fetch reviews or include a free export of another app’s reviews. If you paste external text into spreadsheet software, import it as text rather than allowing it to run as a formula.
If the reviews are still only in a store console, do the export before this step. Exporting App Store and Google Play reviews to CSV covers the seven columns you get, the formatting you should not fix by hand, and the parts of a review the file never contains.
Limits
A filled worksheet is still a worksheet.
The file is where the evidence lives. It is not a verdict, and it cannot speak for people who did not write a review.
- Rows are not usersOne review is one public account, written by someone who chose to write it.
- Counts are sample countsSay “18 reviews in this saved sample”, never a percentage of users.
- Labels can be wrongA working label is a hypothesis with a column of evidence beside it.
- The blank file is the free partComplete reports include excerpts and PDF/CSV exports; the blank worksheet does not.
06 · Optional head start
Start with a report, then check the evidence.
ReviewFold groups feedback from up to 200 recent public reviews of one app in one region. A limited free preview shows themes and counts; complete reports include excerpts and PDF/CSV exports. You can use the report as a starting point and keep your own interpretations in this worksheet.
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.
- The blank worksheet is free and needs no account.
Start with the evidence
Fill the first row before you analyse anything.
Download the blank file, or bring the grouped themes from one app and region and check them against the original sentences.