Eight Days to Submission, Four Known Issues
Submission is in eight days. Your build is going to a platform holder's certification review, and the checklist covers suspend and resume, controller disconnect, save-data integrity, error handling and network-loss behaviour. The candidate has four open issues. (a) If the console is suspended during an online match and resumed, the player returns to a black loading screen indefinitely and must force-quit. (b) The autosave spinner never appears on the first save of a new profile; later saves show it. (c) When the controller disconnects mid-cutscene, the game pauses but the reconnect prompt does not accept input from a second controller, so only a console restart recovers. (d) On network loss during matchmaking the game shows a generic error dialog with only an OK button that returns to the same failing screen. The producer wants to ship all four as known issues and patch later. First-pass review historically takes five business days, and a failed submission costs roughly two weeks of slot time.
1. Classify each of the four issues by certification risk and recommend fix, waiver or defer, with your reasoning.
Read the worked answer
Classify against the current published checklist - verify each line against the live document rather than memory; the categories below are the common shape, not verbatim clauses.
- (a) Suspend/resume into a black screen - blocking. Requirements generally demand that a suspended title resumes to a usable state or exits cleanly. A hang needing a force-quit is exactly the item that fails a first pass and is not waivable. Fix before submission.
- (b) Missing autosave indicator on the first save - likely blocking; check the wording. Save-progress communication and save integrity appear on most checklists. Intermittent is worse than consistently absent, because it looks like a race in a data path. Treat as fix-now unless the clause clearly permits it.
- (c) Disconnect prompt ignores a second controller - high risk, probably blocking. Disconnect handling must let the player recover without restarting the console. Fix, or get explicit platform guidance for that exact scenario.
- (d) Network-loss dialog loops back to the failing screen - medium-high. Error handling usually expects a clear message and a route back to a stable state such as the title screen. It is a dialog-flow change, measurable in hours against a two-week slot loss.
Recommendation Fix all four. None is a genuine quality judgement - they are pass/fail checklist items where the failure mode is losing the submission slot. If any were truly ambiguous, the move is a pre-submission question to platform developer support with a repro video, not a gamble on a reviewer's interpretation.
2. The producer says certification is a formality and quality is what matters. Explain the difference between cert compliance and product quality, what a failed submission actually costs, and the decision process you would run in the next eight days.
Read the worked answer
Two different gates Certification is a defined list of observable behaviours that reviewers execute. It says nothing about whether the game is fun, balanced or polished. A build can be cert-clean and a bad launch; a great game can fail cert on a dialog box. Both matter, but they are decided in different rooms with different evidence, and neither substitutes for the other.
What a failed submission really costs Not just two weeks of slot: the re-test queue position, marketing and store-date dependencies, the team's context switch back into code they thought was frozen, and the risk of a second failure on a defect introduced by the emergency fixes. Hours to fix versus weeks to recover - that asymmetry is the whole argument, and it is a business argument, not a QA preference.
Process for the eight days
- Map every open issue to a checklist line with the reviewer's likely repro. If it cannot be mapped, it is a quality issue and goes through normal severity.
- For each mapped item, decide fix, waiver request, or written risk acceptance signed by the producer. No silent acceptance.
- Run a mock certification pass ourselves: clean profile, retail mode, no dev menus, no unlocked content - the most common failure cause is a build QA only ever tested with cheats on.
- Freeze scope: fixes only, each with a targeted regression, and a go/no-go checkpoint 48 hours before submission with the residual risk written down.
3. Focus on the autosave spinner. Is it actually certification-blocking, what evidence do you need, and how hard should QA push for the fix?
Read the worked answer
Is it blocking? I would not assert that from memory, and neither should the producer. First action: read the current checklist wording on save-data handling and progress indication, and if it is ambiguous, ask platform support with a 30-second repro video. A wrong guess costs either a wasted fix or a failed submission; the question takes an hour.
Evidence to gather
- Exact repro: new profile, first autosave trigger, no spinner; second save shows it. Is it the first save ever, the first save per session, or a race where the save finishes before the spinner is drawn?
- Does the save actually complete? Verify the file on disk and run a power-loss test mid-save. A missing indicator on a successful save is a different defect from a missing indicator masking a slow or failing save.
- Frame-level capture to distinguish late from absent.
- Does it reproduce on every SKU and in retail mode without dev overlays?
How hard to push The producer's framing - ship it, it is only a spinner - is only reasonable if we know the save is reliable. We do not know that yet, and the symptom is in the one system whose failure mode is player data loss. Push for the fix on the evidence: it is likely a one-line ordering issue, the repro is deterministic, and the downside is corrupted saves on launch week. If the producer still declines, put the risk and the unknowns in writing so the decision is explicit and owned. That is the professional line: not demanding, but not silent either.