Skip to content

Is There a Chrome Web Store API for Install Counts? Yes, in V2

6 min readModerok team

The Chrome Web Store API V2 returns install counts through publishers.items.queryMetrics. The exact call, every metric it supports, and its limits.

Yes. The Chrome Web Store API has a documented metrics method, publishers.items.queryMetrics, which returns installs, uninstalls, users and impressions for one of your items as daily rows. It is a GET call and it works with a read-only OAuth scope, per the V2 REST reference as of November 2026. If you had concluded no such API existed, that is understandable. The archived V1 reference documents an Item resource of kind, id, publicKey, uploadState and itemError[], with no metric anywhere, and the blog post that introduced V2 on 15 October 2025 does not mention queryMetrics.

TL;DR: GET https://chromewebstore.googleapis.com/v2/publishers/PUBLISHER_ID/items/EXTENSION_ID:queryMetrics with metrics=installs and a timeSpec date range returns install counts, no dashboard clicking and no CSV download. It covers store-side aggregates only, so what happens inside your extension still has to come from your own instrumentation.

The call that returns install counts

The documented endpoint is GET https://chromewebstore.googleapis.com/v2/{name=publishers/*/items/*}:queryMetrics, described in the reference as "Query the metrics for an item." Two query parameters are required, metrics[] and timeSpec, and the request body "must be empty." timeSpec holds an optional firstDate and lastDate, each a Date of year, month and day, both bounds inclusive. The reference prints no example request, but the API's own discovery document (revision 20261111, retrieved 13 November 2026) lists the flattened query parameters, so the spelling is not a guess:

curl -G -H "Authorization: Bearer $TOKEN" \
  "https://chromewebstore.googleapis.com/v2/publishers/PUBLISHER_ID/items/EXTENSION_ID:queryMetrics" \
  -d "metrics=installs" \
  -d "metrics=uninstalls" \
  -d "timeSpec.firstDate.year=2026" \
  -d "timeSpec.firstDate.month=11" \
  -d "timeSpec.firstDate.day=1" \
  -d "timeSpec.lastDate.year=2026" \
  -d "timeSpec.lastDate.month=11" \
  -d "timeSpec.lastDate.day=13"

metrics and dimensions are both repeated parameters, so you pass each value as its own metrics= pair rather than a comma-separated list. The response is metricRows[] plus an optional nextPageToken. Each MetricRow has a date, a dimensionValues[] column set and metricValues[], and each MetricValue carries metric, value and, where it applies, uniqueUserCount. Paging defaults to 100 rows, the maximum pageSize is 1000, and larger values are coerced down.

Every metric queryMetrics supports

All seven, quoted from the reference as of November 2026:

MetricDocumented as
installs"Count of item installs."
installs-7d"Count of item installs in the last 7 days."
uninstalls"Count of item uninstalls."
users"Count of daily active users of the item."
users-7d"Count of 7-day active users of the item."
impressions"Count of item impressions across the entire store front."
page-views"Count of individual views of the item's detail page."

Those last two overlap more than the names suggest. The dashboard's metrics page says "An impression occurs when your extension is featured in any collection or direct visits to your store item", so direct visits sit on both sides of the line. Do not read impressions as pure discovery and page-views as pure listing traffic.

One dimension per request

dimensions[] is optional and the reference is explicit that "At most one dimension can be specified." The documented set is os, app-version, country, install-source, install-type, language, enabled, and the UTM group: utm-campaign, utm-content, utm-medium, utm-source, utm-term.

A few of those answer questions that previously meant squinting at the dashboard. install-type is documented as "The type of install (from webstore, inline, or other)", which is the split that makes store totals and your own numbers diverge. app-version crossed with users gives you version spread over time, which beats guessing from the release date. The utm-* dimensions are documented only as the UTM campaign, content, medium, source or term "of the user", and Google's post on revamping the dashboard analytics notes that publishers using UTM parameters for traffic attribution see listing page view data broken down by utm_source, utm_medium and utm_campaign.

One dimension at a time means a per-country-per-version breakdown costs several calls and a join on your side.

What you need to call it

Per the Chrome Web Store API setup docs, queryMetrics is one of the read-only methods, so https://www.googleapis.com/auth/chromewebstore.readonly is enough and you do not have to hand a reporting script a scope that can publish. Your publisher ID "is displayed in the Publisher > Settings section" of the dashboard, and the extension ID is the one in the listing URL. Credentials come from a Google Cloud OAuth client or from a service account, which the API overview says V2 supports, noting that "V2 uses the same OAuth 2.0 scopes as V1" and that V1 was the latest version only "Prior to October 2025". The announcement put V1 support at "until 15th October 2026", a date that has now passed.

What queryMetrics still cannot tell you

It is store-side telemetry, and the seven metrics are the whole of it. Three of them do describe life after the install (uninstalls, users, users-7d), but that is as far inside your extension as the store can see. There is no custom event, no feature usage, no error count, no activation rate and no retention by install cohort in that list, because the store is not running your code.

The two active-user metrics also come with their definitions sealed. users is documented as "Count of daily active users of the item." and nothing on the reference says what makes a user active that day, how profiles are deduplicated, or how the figure relates to the separate Users report in the developer dashboard, which has its own definition. We pulled the dashboard's numbers apart in Chrome Web Store Weekly Users vs. Installs.

The reference says nothing at all about freshness, lag or how far back the data goes, so treat all three as unknown. If you build a daily job on this, make it tolerant of a day that comes back empty or thin instead of paging you about it.

Your own numbers will still disagree with it

If you count installs with chrome.runtime.onInstalled as well, expect two different numbers, and do not treat either as the broken one. The fullest account of the store's side is still a February 2019 chromium-extensions thread on programmatic install analytics versus Chrome Web Store analytics, where a poster identifying themselves as "an owner of the system that generates the counts Chrome Web Store reports" wrote that "at time of writing" the counts were user-driven install attempts from the Chrome Web Store, that "right now they only reflect web-store install methods", that installs arriving on a user's other devices through Chrome Sync do not add to them, and that they "are deduplicated by user, but only on a daily basis." Those are hedged, seven-year-old statements and the V2 reference documents none of this, so treat them as the best available explanation rather than the current spec. Your own onInstalled listener, by contrast, fires in every profile that runs your worker, including deployments the store never sold.

Managed ChromeOS profiles are the sharpest version of the gap. Google's ephemeral mode documentation says such a profile "is marked for deletion only after the user signs out or manually closes every window associated with the profile" and "is deleted the next time Chrome starts", so the extension lands on a fresh profile at the next sign-in. Moderok's docs describe the consequence, onInstalled firing with reason: "install" again for the same person on the same device, in why its install count differs from Chrome Web Store, along with what the dashboard filters out because of it. Chrome's own documentation does not spell this case out. The counting rules behind each number are most of the answer to how many users your extension has.

Instrument the half the store cannot see

queryMetrics is good news: the store numbers are now scriptable, and a nightly job with a read-only scope can keep them in your own warehouse. What it will never return is product data, because the store does not see your extension run.

That part is what Moderok is for: an analytics SDK for MV3 service workers, 5.7 kB gzipped, zero runtime dependencies, no DOM globals in its source and no remotely hosted code. Setup is three parts, npm install @moderok/sdk, the storage permission, and Moderok.init({ appKey }), after which __install, __update, __daily_ping and __error arrive without a track() call, while your own Moderok.track() events cover the features you want to judge.

Start with the getting started guide, check the automatic events reference for exactly what gets sent, or see what the dashboard shows first.