The Review Atlas

Look at the 1,800,287 reviews in this dataset and you'll find 100 percent of them come from the United States. That's not a dominant share of some global total. It's the whole thing. Whoever put this data together either filtered everything down to the US or simply never had reviews from anywhere else to begin with, and that changes what you can honestly claim from the rest of these numbers. A rating average, an app count, a per capita rate, none of it tells you how the US stacks up against other countries, since there's no other country in the data to measure it against. Treat this as a US review ledger, not a world map. Before anyone builds a global narrative on top of it, ask the data vendor straight out whether other countries ever appear in their raw feed.

In the US, the average star rating comes out to 3.4. That's well below the 4 and 5 star averages most App Store dashboards like to flash, and since there's no other country here to measure it against, it's the only rating figure we've got. An average like this masks the two groups who actually bother leaving reviews: people who are furious and people who are thrilled, while everyone in between just stays quiet. A 3.4, sitting below the midpoint on a five point scale, suggests the angry reviewers in this US sample aren't being fully offset by the happy ones. Without a second country's average to set beside it, there's no way to tell whether 3.4 says something specific about the US or just reflects how App Store reviewing tends to go everywhere, and that missing comparison tells you more than the number itself.

Leave the country question to one side for a moment, because the breadth number on its own tells you how far this dataset actually reaches: 14,479 distinct apps picked up at least one US review, out of 1,800,287 reviews total. That's a wide net across the App Store catalog, not a handful of blockbuster titles getting reviewed over and over again. For a publication built on this data, that spread is what makes any later app-level cut worth trusting. You're not drawing conclusions from three mega-apps here, you're pulling from a catalog scattered across thousands of listings. The figure shows up as 100 percent of the 14,479 in its own cut, which is just a reminder that this number, like the country figure before it, has no second population inside this dataset to stack up against.

The US puts up 5,267.3 reviews per million people in this dataset. That number holds up fine as a measure of intensity on its own, since it's capturing how review-prone the sampled population is relative to its size rather than just a raw headcount. But a per capita rate only really proves useful once you can hold it up against something else, and this study has exactly one country on the chart. There's no second country's rate sitting next to that 5,267.3 per million, so there's no way to say whether that's high, average, or low for a connected, smartphone-saturated market. Think of this figure as a baseline still waiting for a comparison the dataset simply doesn't offer, not as some verdict on how Americans review things.

This cut came back with no tiers and no values, which is a real problem given what the study was actually trying to answer: does review volume track with internet access. The internet tier cut has zero rows, and the country cut only contains the US, so there is no way inside this dataset to test population size or connectivity against where the reviews actually come from. That kind of comparison needs multiple countries tagged by access tier, and this sample just was not built that way. If you are building on top of this feed, ask for internet tier tagging explicitly at ingestion. It does not show up by default, and a blank cut like this one can derail an entire analysis if nobody catches it first.

Of the 1,800,287 reviews in the dataset, 1,212,796 of them, or 67.4 percent, landed in 2026 alone. Throw in 2025's 233,037 reviews (12.9 percent) and you're looking at roughly four out of every five reviews on record coming from just the last two years. Compare that to 2008, which opened the series with a mere 457 reviews (under 0.1 percent of the total), or even 2013's relative high point of 3,205 reviews at 0.2 percent. Nothing gradual about this growth. It shoots almost straight up right at the end of the timeline, and that kind of shape usually points to one of two things: either review activity genuinely exploded in recent years, or the collection window is backloaded and researchers just didn't sample the earlier years as thoroughly as the reviewing happening in real time. Either way, any trend line drawn across this eighteen year span is going to be dominated by 2025 and 2026. The pre-2019 years, each sitting at 0.9 percent or less on its own, deserve to be treated as thin, low-volume context, not a stable historical baseline.