Spam and rating recovery

Spam moderation, Spam Automation and Rating Recovery are paid features: they need Starter or above. On Free, the Spam folder and these pages are locked.

Every review is checked for spam signals as it arrives: links, email addresses and phone numbers, promotional, crypto or prize scam wording, disguised text (hidden characters, look-alike letters, stylized fonts), and the same text repeated across reviews. The AI also flags reviews, including abusive ones. Very short one- and two-star reviews with no real feedback can be marked as suspected, but never confirmed automatically.

Flagged reviews stay in the inbox with a Likely spam marker and a "Why flagged" box listing the signals, and they also appear in the Spam folder. They are not hidden. A classifier that quietly buries real customer complaints would be worse than no classifier, so the call stays yours. If you mark a review as not spam, that text is never flagged again.

Nothing flagged as spam or abusive is ever auto-replied to, whatever your automation rules say.

Reporting a review hands it to the platform for moderation. The platform decides. Nobody (us, you, any tool you can buy) can delete a review from a store, and the industry is full of people implying otherwise.

None of the live platforms offers a moderation API, so reports are filed on the platform's own site. The report button takes you there:

Platform	Where the button takes you	How to report there
Apple App Store	The app's Ratings and Reviews page in App Store Connect	Report a Concern, reason Spam
Google Play	The Play Console: the exact review if you linked the app's Play Console URL, otherwise the Console home page	Report the review and confirm
Google Business Profile	A Google search for your business	Report review from the review's menu, reason Spam

After you file it, the review is marked Reported, and we keep checking the store to see whether it was removed. You can see which reports ended in a removal.

Facebook, Instagram and TikTok are coming soon, so nothing from them is reported today.

Filing reports one by one is slow. Spam Automation lets your own AI assistant file them for you, on a schedule.

The stores only accept spam reports from their own sites, so the assistant files them in a browser where you are already signed in to the store. ScaleSensor never receives your store passwords, and the assistant is told never to enter passwords or solve CAPTCHAs.

Connect one of these on the Setup tab:

• Claude: Claude Desktop or claude.ai, adding AI Review Sensor as a custom connector, with the Claude in Chrome extension for the browser.
• Codex: the Codex CLI, adding the connector from the command line, with the ChatGPT Chrome extension for the browser.
• Any remote connector: any assistant that supports remote MCP connectors.

The Setup tab gives you the connector URL and the exact steps for each. You can also press Report now to run a batch whenever you like. The assistant works in batches of 15 reviews by default (up to 25).

Set how often reporting runs: Every day, Every few days (every 2, 3, 4, 7 or 14 days) or Every week on a day you choose, at a time in 30-minute steps. The schedule uses your workspace timezone, set in ScaleSensor Settings.

The schedule runs as a scheduled task inside the Claude desktop app or the Codex app (one runner per workspace). You set it up by pasting one setup prompt into that app. We cannot start your assistant from our side, so the task only runs while the app is open; if it was closed at the scheduled time, it runs when the app next opens.

To stop, choose Turn off. Scheduled reporting stops, and you can still report any time with Report now. You are then offered a prompt that pauses the task in your app rather than deleting it.

The Activity tab shows every run and why it ended. Unfinished reviews always go back in the queue for the next run. The reasons you may see:

Reason	What to do
The store console was signed out	Sign in to the store in that browser, then run it again
The store account cannot report reviews for this app	Check the store account's access, then run it again
The store limited how fast reports could go out	Nothing. The rest are offered on the next run
The store could not be reached	Nothing. The rest are offered on the next run
A store page could not be found	Nothing. The rest are offered on the next run
The browser was not available	Open Chrome with your assistant's extension connected, then run it again
Stopped at your request, or the assistant stopped before finishing	Run it again when ready

A review that fails is held back for 6 hours before it is tried again, and after 3 failures it is set aside as needing your attention.

Missed runs. If a scheduled time passes and nothing reaches AI Review Sensor (the app was closed, the computer was asleep, or the connector was not available), the run is marked Missed. Nothing was reported, and the spam is kept for the next run. To catch up, open the task in Claude or Codex and run it now.

Outdated task. If you change the schedule, the task in your app still runs the old time until you update it. A banner tells you when that is the case and who needs to update their task.

The Rating insight page answers one question: what would your rating look like without the reviews you believe are fake?

It shows your recent rating against your all-time average, the rating recovered from spam the stores have already removed, the rating you can recover if the flagged spam still on the store were removed, and how much spam is awaiting a report, awaiting removal, or removed. That gap is the size of the prize, and quite often it is smaller than people expect, which is worth knowing before you spend a week filing reports.

You also get a trend chart with releases marked, sentiment, What's moving your rating (the topics and versions behind it), rating by version for a single app, and a short list of actions. Filter by platform, app and version.

Two invariants, and they are deliberate:

Spam never changes a real rating. Your rating includes flagged spam until the platform removes it. The "without spam" figure is shown beside your real one, never in place of it. Your public rating is what the store says it is.

We only ever claim what actually happened. A recovered rating is counted once a platform has genuinely removed a review (we check by fetching that exact review again), not when you flagged it and not when you reported it. The steps are always Flagged, Reported, Platform removes, Rating recalculates.

This matters legally as well as ethically. Manipulating review content or presenting adjusted ratings as real is exactly what regulators have been fining companies for. The feature exists to find genuine spam, not to make your rating look better than it is.