Apple's notarization reviewers enforced tracking and login rules on a rating app distributed outside the App Store

Apple's own description of notarization already goes past a malware scan. Apple calls it a baseline review focused on platform security, privacy and device integrity ^[5]. Automated checks plus human review confirm that an app is free of known malware, functions as promised and does not expose users to egregious fraud ^[5]. Builds headed for alternative marketplaces in the EU, Japan and Brazil go there instead of App Review ^[4]. Rate Me ships through AltStore PAL, and its developers have published their record on dev.to: six notarization rounds, two of them rejections ^[1][2].

The first rejection was a consistency check. A reviewer opened the app on an 11-inch iPad Pro running iPadOS 26.6.1 and looked for the tracking prompt ^[9]. The team had removed ATT in July, and the code contains no ATTrackingManager ^[10]. The Info.plist still declared NSUserTrackingUsageDescription. The key came back the way stale config usually does, copied from an old working copy on another machine without checking history ^[10]. A declared tracking intent with no prompt on screen failed Guideline 2.1 ^[9]. App Review had caught the mirror case two months earlier, with the key and a wrapper present and nothing calling it ^[11]. The developers now ship the key only when code calls ATTrackingManager.requestTrackingAuthorization, and they keep usage strings in English because review runs on an English system ^[11][18].

The second rejection reached past the binary. A different reviewer, on an iPad Air M3, opened the sign-in page at auth.rateme.lv, found Google and Facebook buttons, and cited Guideline 4.8 Login Services ^[12]. Those buttons are on a web page the app opens, not in the app itself ^[13]. The fix is the part I would copy. The team added Sign in with Apple on the identity service side, with an App ID and capability, a Services ID, a key and a return URL ^[14]. They resubmitted the same build 8 with a reply to the reviewer and were approved a bit over a day later ^[14].

Content is where the review stopped. Rate Me has people upload photos and collect 1-10 ratings from other people ^[3]. App Review closed the App Store to it under Guideline 1.2, for rating people on their appearance, and also cited 5.1.1(v) for missing in-app account deletion ^[6]. The unchanged mechanic then passed notarization four times, and neither notarization rejection mentioned 1.2 or content ^[7]. The write-up does not say whether account deletion was added before notarization. On this evidence, a developer leaving the App Store over a content rejection should plan for a human who tests declared behaviour and linked pages. The evidence does not show that human judging the mechanic, and it covers one app. The developers wrote that "alternative distribution is not a loophole, it is a different kind of review." ^[8]

The queue figures are one account's timeline. Version 1.0.3 took a week end to end: review started within three hours, and the verdict took seven days ^[15]. The next release, 1.0.4, took 33 hours, and the 1.0.5 rounds ran 8 to 42 hours ^[15]. By 1.1.0 it was 3 hours 46 minutes, with a one-minute review ^[15]. Counting the week as 168 hours, turnaround fell about 45-fold ^[19]. The developers say they cannot tell whether Apple's queue warmed up or their account built a track record ^[16]. That number transfers only if the cause is the queue. I would plan a first submission around the week. Two rejections in a row still cost three days on 1.0.5 ^[17].