At 00:59 on November 4, App Review rejected TempoTrack 1.0 (build 3) under Guidelines 4.3(a), 4.3(b) and 4.2.6. The build had gone into the queue on October 31 at 14:41. Four days of waiting, three clauses in the reply.
The letter is the standard 4.3 template: the app "shares a similar binary, metadata, and/or concept as apps submitted to the App Store by other developers." 4.3(a) is the multiple-Bundle-ID clause. 4.3(b) is the saturated-category clause. 4.2.6 is the commercialised-template-or-generation-service clause. TempoTrack is an intonation coach: you play a passage into the microphone, and afterwards it tells you which notes were off and by how many cents.
The letter also contained an instruction that turned out to be a dead end: if I believed the app complied, I should reply to the message instead of resubmitting. So I replied. At 21:40 on November 7 I pasted 3,897 characters into the Resolution Center box, which caps at 4,000. I know the count to the character because I measured it before pasting — the box caps at 4,000 and I wasn't going to discover its truncation behaviour with a finished argument:
text = open("appeal.txt").read()
print(len(text)) # 3897
print(4000 - len(text)) # 103 characters of headroom
Then I waited. Apple's own FAQ says messages in Resolution Center generally get a response within 24 hours. At 20:45 on November 10 — 71 hours and 5 minutes later — there was still nothing. Not a rejection, not a request for more information, not an acknowledgement.
That silence is the real subject here, because it taught me something the rejection itself didn't: I had chosen the wrong escalation path, and the UI gave me no way to see that.
What went into the 3,897 characters
The rejection asked nine numbered questions: what the app does, how it differs from existing tools, which frameworks it uses, whether it was generated from a template, whether it had been beta tested, and so on. I answered all nine, in order, because a partial answer to a spam rejection reads like evasion.
The two answers I spent the most characters on were the two that a 4.3 letter actually turns on.
"Different from existing tools" is not an argument, it's a category claim. So I named the three groups of tools that already exist and what each one structurally cannot do. Tuners show you the current pitch and keep no record. Recording apps let you play a take back, but judging intonation by ear requires training most musicians don't have. Practice logs record duration and nothing else. None of them tells you, after a session, which notes were flat and by how much. That gap — after-the-fact, per-note diagnosis in cents — is the whole product. If a reviewer can't see the gap in your first paragraph, the rest of the appeal is decoration.
4.2.6 alleges a template or a generation service, which is a claim about the artifact. So I answered it with a measurement from the repo:
20 source files
~8,300 lines total
~2,400 lines of pitch detection (McLeod method) alone
A template does not carry 2,400 lines of signal processing, and I said so in exactly those terms. That number is the only part of the appeal that a reviewer can check against the binary they already have, which is why it survived every later edit.
The honest answer that probably hurt. One of the questions asks about beta testing. I had zero TestFlight groups. Not "a small group" — zero, never created, because the app had no users yet and I'd been treating external testing as a launch-week activity rather than a review artefact. I wrote that down plainly instead of dressing it up. A build with no external testers is, from the outside, indistinguishable from a build nobody has ever run, and that's a bad shape to be in when the accusation is "this is a generated app."
71 hours, and the table I should have had in front of me
Here is the part I got wrong, and it wasn't the writing.
I assumed "reply" and "appeal" were two names for the same escalation, with reply being the cheaper one. They aren't. Reply is a message in a thread that a reviewer may or may not read. Appeal goes to the App Review Board and has an actual process attached to it. The rejection letter only tells you to reply, so reply is what you do, and then you sit in a channel that has no obligation to answer.
The behaviour is predictable once you know the rejection type:
| Rejection type | What Apple actually does |
|---|---|
| 2.1 Information Needed | Replies, because they are waiting on you |
| 2.3.7 Metadata | Replies, because they need to confirm the fix |
| 4.3 Spam (a verdict) | Usually silent. The conclusion is already reached |
4.3 is a judgement about what your app is, not a list of missing materials. When you send an argument rather than an artefact, and the argument doesn't land, the default action is nothing. Not a rebuttal — nothing. I spent 71 hours reading that nothing as "still being considered."
The cost was concrete: three days of a stalled release, plus the thing I couldn't undo — I had spent my one clean shot at the thread on a message nobody was required to answer.
The box that holds 2,000 characters
The appeal form has a field called "Reason for appeal." It is capped at 2,000 characters.
My draft was about 3,200. That's not a formatting problem, it's a triage problem: the form forces you to decide what your case actually rests on, and it does it after you've already written the comfortable version.
I got it to 1,993 — seven characters of headroom — by cutting every sentence that restated something already said. Gone: "these are not variations of one product; they are different products that happen to share an owner," which is a nice line that adds nothing after the previous paragraph has already listed five text-entry apps and one audio app. Gone: the two separate reassurances that I wasn't disputing the process. Kept, unconditionally: the three numbers (20 files, 8,300 lines, 2,400 lines of DSP), the three-way split of existing tools, and the closing request that Apple name the app mine supposedly duplicates — because if they can't name it, that's the strongest fact in the whole document.
The compression was the most useful hour of the entire episode. The 3,200-character version was persuasive to me. The 1,993-character version is checkable by someone with thirty seconds and the binary.
What I gave up
I did not resubmit a build. A fresh build restarts algorithmic review and sometimes clears a 4.3 on its own — that's the fast path, and it was available. I gave it up because resubmitting closes the message thread, and closing the thread while an appeal is pending is the one move that can't be walked back. I chose a slow, uncertain path over a fast one that would have destroyed the record.
I chose "App Rejection" as the appeal topic, not "Unfair Treatment." Unfair treatment is the emotionally correct option and the strategically wrong one: it converts a disagreement about how a guideline was read into an accusation that the reviewer was biased. I have no evidence of bias, only of a bad fit between a template rejection and a real app. Losing the moral high ground was cheaper than losing the appeal.
I wrote one appeal, knowing I get one. Apple allows a single appeal per rejection. That means no iteration, no "let me try a different framing next week." Every later idea I have about this rejection has to go into the same document or nowhere, which is why the three numbers are in it and the adjectives aren't.
And I stopped treating metadata as leverage. No new screenshots, no renamed subtitle, no icon swap during the appeal window. Those are things you do to a listing; they don't answer a question about whether the binary is original.
What I'd do differently
Two things, both cheap.
Put a screen recording in the first reply, not the appeal. The strongest version of my case is fifteen seconds of someone playing a phrase badly and getting a per-note report. That's a demonstration of the gap; the 3,897 characters are only a description of it. Apple's reviewers are looking at thousands of apps a week and a video survives compression better than prose.
And answer 4.3 with the repository, not with the category. "This category is crowded but here is the hole" invites an argument. "Here are 2,400 lines of pitch detection in a build that shares no code with my other five apps" ends one.
The release is still stalled. That's the honest state of it.
Question
If you've been rejected under 4.3, what actually moved the reviewer — a new build, a screen recording, or the appeal itself? I'm specifically interested in whether anyone has seen a Resolution Center reply produce a reversal, because across every account I've read, the reversal happens after a build change or an appeal, never after an argument. If you have a counter-example, I want to hear it before I spend my one appeal.
Top comments (0)