A campaign is not operationally ready because the final files were delivered. It is ready when at least one creative can serve in the intended account, market and line item—and another approved option can take over while the next change is reviewed.
That distinction matters to global brands coordinating product teams, Chinese suppliers, production partners and media buyers across time zones. A late URL correction or tracking change can turn a finished campaign into a waiting queue. The solution is not to demand an impossible promise that every platform will approve on schedule. It is to procure servable creative inventory and release updates without removing every known-good option at once.
Platform review is a deployment dependency
The rules differ by platform, so a single review-time assumption is unsafe.
Google Display & Video 360 says any update to creative content or a landing page restarts review, including asset, landing-page URL, tracking URL and appended HTML changes. Its warning is unusually concrete: if every creative assigned to a line item is changed, that line item may have no servable creative for eight hours or longer. Google also documents a draft workflow, but publishing the draft replaces the published version and pauses it until approval.
TikTok’s Ad Review FAQ, last updated in February 2025, says most reviews typically complete within 24 hours and that editing creative, links or certain targeting fields can trigger a new review. Its more detailed change table explains an important difference: in an example involving a URL edit, the approved ad group can remain active during review; the update takes effect after approval, while a failed review can pause the ad group. Do not translate a Google-specific outage rule into a TikTok guarantee—or vice versa.
Amazon DSP’s review guidance, updated August 11, 2026, says reviews typically complete within 12 hours but may take longer around Prime Day and Black Friday/Cyber Monday. Re-saving an ad restarts review, and attaching a new ad group can change the approval outcome because ad-group settings are part of eligibility.
These are documented platform rules. The operational conclusion below is ONCE’s analysis: approval should be managed as inventory state, not treated as a one-time box checked at delivery.
Replace the final folder with three inventory states
The model has three bins. Each record is specific to a platform, account, market and dependency set.
Live inventory
These creatives are approved and genuinely eligible to serve now. Their destination, tracking, product availability and rights must still be valid. An approved file pointing to a broken or expired destination is not live capacity.
Warm standby
These are already approved alternatives that are paused or carrying little traffic. They can assume spend if the live version is edited, rejected or retired. A finished export waiting to be uploaded is not warm standby; it is only a candidate.
The standby should also avoid the same failure domain. If both versions depend on one expiring offer page, the account still has a single point of failure. A good reserve has a valid claim, current rights, a working destination and enough creative difference to keep serving its intended role.
Candidate inventory
These versions are still in production, internal review, platform review or controlled testing. Candidates are useful, but a media plan must not count them as confirmed capacity. Record the blocking state explicitly: claim approval, talent rights, localization, landing-page readiness, upload QA or platform review.
For each asset, keep six fields: platform/account, creative ID and version, approval status with timestamp, destination/tracking/product dependencies, planned activation window, and rollback target. That ledger can extend an existing campaign version-management process; it should not become another disconnected spreadsheet.
Use a blue-green release for creative changes
Blue-green is an operations analogy, not a feature name used by the ad platforms.
- Freeze blue. Protect the current approved creative, its copy, destination and tracking from in-place edits.
- Create green. Build the update as a separate creative with its own ID.
- Review in context. Submit it in the intended account, market and ad-group conditions; confirm the destination and tracking work.
- Shift deliberately. Move traffic only after green is eligible to serve and passes the dependency check.
- Retain rollback. Keep blue available for a defined window before retirement. Never edit the whole live set simultaneously.
Some platforms may allow the previous state to keep serving while an edit is reviewed. That can shorten the overlap period, but it does not remove the need to know which state is live or what happens after rejection. For fixed-date launches, regulated categories and major retail events, submit early: the platforms describe typical review times, not guaranteed service levels.
Buy continuity, not just file count
A normal production quote divides production and localization expense by the number of delivered assets. That can reward output while hiding upload QA, resubmission work and outage exposure.
Add a second, operational denominator:
Continuity-adjusted creative cost = (production + localization + upload QA + resubmission work + reserve maintenance) ÷ approved creative-days during the planned flight
One creative eligible to serve in the target account and market for one day equals one approved creative-day. A delivered asset still pending review contributes zero. This is not a performance metric and should not replace incrementality, ROAS or conversion quality. It measures the cost of preserving deployable choice.
Teams can also track a continuity coverage ratio: actual approved creative-days divided by the approved creative-days required by the media plan. There is no universal target. A one-day organic post needs little or no reserve; a global product launch, seasonal promotion or always-on acquisition program needs more.
When a standby is worth buying
Specify warm standby in the production scope when the campaign has a fixed launch, continuous spend, frequent localization, multiple markets, regularly changing landing pages or a high cost of interruption. Skip the full system when content is organic, timing is flexible and there is no paid-media dependency.
The wasteful version of redundancy is a folder of near-identical copies that fail together. The useful version is an independently viable creative with valid claims, rights and destinations, already approved in the conditions where it may need to take over.
For the next brief, ask the production and media teams to deliver three connected artifacts: a version matrix, a platform approval ledger and a blue-green change plan. ONCE’s overseas video production service can use that operating model to plan alternative hooks, product proof and market variants before the shoot. The procurement unit is no longer “a batch of finished videos.” It is enough approved capacity to keep the campaign deployable while the next version changes.
