Every public company that ships a consumer app leaves a trail that updates daily. It is not in a filing. It sits in the review section of the App Store and Google Play. Anyone can read it. Few people read it correctly.
This article explains how to treat app review data as a demand signal, why the rate of change matters more than the headline numbers, and where the data quietly lies to you. It is not investment advice.
What the review section actually measures
An app listing shows two things that look like demand but are not the same thing.
The first is the aggregate rating, usually a star average and a total review count. The total count is cumulative. It grows forever unless the developer resets it with a new version, which Apple allows. A high total tells you an app has been around and has been used. It tells you nothing about this quarter.
The second is the stream of individual reviews, each with a date, a star rating, a version number, and text. This stream is the raw material. Everything useful comes from reading it as a time series rather than as a snapshot.
Think of the difference the way you would think about a balance sheet versus an income statement. The total count is a stock. The daily flow of new reviews is a flow. Demand lives in the flow.
Velocity: the number that moves
Rating velocity is the count of new reviews per day or per week. It is the simplest derived metric and the most informative one.
Why does it work? Only a small fraction of users ever write a review. That fraction is roughly stable for a given app over short windows, because the prompt that asks for a review is built into the app and fires on a fixed logic. If the share of users who review stays flat, then a change in review count is a change in active users.
That makes velocity a proxy for engagement, not for downloads. A user who installs and never opens the app does not get asked to rate it. A user who opens it daily eventually does. So velocity tracks the part of the funnel that matters for revenue.
To use it well you need three things:
- A consistent collection schedule, so that gaps in your scrape do not look like gaps in demand.
- The app version attached to each review, so that you can separate a product change from a marketing push.
- A long enough baseline to know what normal looks like for that app. Weekly seasonality is real. Many consumer apps see more reviews on weekends.
Once you have those, the questions get interesting. Did velocity rise before the company reported a strong quarter? Did it fall after a pricing change? Did a competitor's launch show up as a dip? None of these are proofs. They are hypotheses you can check against the company's own disclosures.
Volume: the number that misleads
The total review count and the star average are what most dashboards show. They are also the numbers most likely to send you the wrong way.
The star average is a lagging indicator by construction. If an app has tens of thousands of ratings at 4.7 stars, a bad release that generates a thousand one-star reviews barely moves the average. The damage is real. The headline hides it. You would see it instantly in the velocity of one-star reviews, which is why you should track velocity by star bucket, not just in aggregate.
The total count is worse. It mixes years of history into one figure, and developers can reset it. A reset looks like a collapse in volume. It is a product decision, not a demand event. Google Play does not offer the same reset, so the two stores can disagree about the same app for reasons that have nothing to do with users.
Where review data lies
There are at least five ways review data will mislead a careful reader. Knowing them is most of the skill.
The review prompt changes. The developer controls when and how often the app asks for a rating. If a release moves the prompt from the tenth session to the second, velocity doubles with no change in users. The fix is to read the release notes alongside the data and to watch for step changes that align exactly with a version bump. A real demand shift ramps. A prompt change jumps.
Incentivised and fake reviews. Both stores prohibit paid or fake reviews, and in the United States the Federal Trade Commission adopted a rule in 2024 that bans fake and incentivised consumer reviews outright. The rule is codified at 16 CFR Part 465. Enforcement exists, but detection in your own data still falls to you. Signs include bursts of short five-star reviews with similar phrasing, reviews that cite features the app does not have, and velocity spikes with no matching change in one-star or three-star volume.
Store policy and ranking changes. When Apple or Google changes how reviews are displayed or which are shown by default, the data you can scrape changes with it. Compare across stores and across time only after checking whether the collection surface itself moved.
Platform and geography skew. iOS users and Android users review at different rates. Users in different countries review at different rates and in different languages. An app that is growing in one region and shrinking in another can show flat global velocity. Always segment by store and by storefront country before drawing a conclusion.
Reviews are not revenue. This is the big one. A free app with a strong review flow may monetise nothing. A paid subscription app may have fewer reviews because its users are fewer and more committed. Velocity tells you about engagement. It does not tell you what engagement is worth. For that you need the company's own numbers.
Tying the signal to the filings
The reason to bother with any of this is that app reviews update daily while official disclosures update quarterly. If the review stream is a leading indicator, the test is whether it predicts what the company later reports.
Public companies describe their consumer products, user metrics, and risk factors in their annual report on Form 10-K and in quarterly reports on Form 10-Q. The SEC publishes the general instructions for Form 10-K, and every filing is searchable through EDGAR full-text search. Searching a phrase like "monthly active users" or "paid subscribers" in that tool, restricted to a single company, gives you the official number the company chose to report. Line that up against your velocity series and you have a real backtest instead of a story.
Institutional investors who hold these companies disclose their positions on Form 13F, with the deadline and the reporting lag explained in our piece on 13F deadlines and the 45-day lag. That lag is exactly why a daily alternative signal is worth having. Review velocity can move months before the next 13F batch shows whether large holders added or trimmed.
Members of Congress who trade these names file under the STOCK Act, and the 45-day rule governs how quickly those trades become public. Again, review data moves faster than the disclosure. The question is whether the two series agree.
Free ways to do this yourself
You do not need a paid dataset to start.
- Apple exposes a public RSS feed per app and per storefront that returns recent reviews as JSON. Google Play has no equivalent public feed, but the store pages are readable and several open source scrapers exist.
- EDGAR full-text search and the company's investor relations page are free and official.
- A spreadsheet is enough for one or two apps. Collect weekly, tag each row with the version, and plot velocity per star bucket.
The limits of the free route show up at scale. Apple's feed returns only recent reviews, so you cannot reconstruct history you did not collect. Multiple storefronts multiply the work. Keeping a consistent daily series for hundreds of apps across two stores is where a maintained dataset starts to earn its cost.
A short checklist
Before you trust a move in review data, ask:
- Is this velocity or volume? Only velocity is a current signal.
- Did a version ship on the same day? Check the release notes for a prompt change.
- Does the move appear in more than one storefront and more than one star bucket?
- Does the text match the stars? A burst of generic five-star reviews is a flag.
- What did the company report last quarter, and what will the next 10-Q let you verify?
If the answers hold up, you have something useful. If they do not, you have noise with a timestamp.
Where to take it next
Review velocity tells you what users are doing this week. The 13F consensus tells you what the largest professional holders did last quarter, once the lag clears. Running the two against each other is the fastest way to find consumer names where the crowd and the institutions disagree. Our Smart-Money 13F Consensus report tracks that institutional side so you can line it up against your own demand data.
Want the signal instead of the raw filings? Get a free report preview. Prefer the tool to the write-up? Browse all data feeds or connect the free MCP server.