The work brought together in the system
Preparing a file, generating a version and publishing it are separate steps. Reel Studio keeps each job's state so a completed preparation can be distinguished from a confirmed publication.
- Receive the file
The upload records the video and the selected preparation options.
- Prepare the versions
The workflow supports editing, titles, captions and variants according to the job's options.
- Review and choose the output
Jobs that need attention are identified for review. The output can follow a queue or the publication configuration.
- Record the publication
The processing state is kept separate from the network receipt, identifier, link and date.
What the records show
The review includes the ZaTek and ZaTek AI installations. Eight jobs marked as tests were excluded, along with source matches and derivatives identified by the method.
| State at the snapshot | Jobs |
|---|---|
| Completed | 55 |
| Needs review | 6 |
| Failed | 5 |
| Cancelled | 4 |
| Total eligible | 70 |
| Outcome | Records |
|---|---|
| Confirmed publications within the period | 28 |
| Failed records | 20 |
| Total linked to the cohort | 48 |
The 28 confirmed publications are on Instagram: 12 for ZaTek and 16 for ZaTek AI. A completed job is not counted as published without its receipt. Variants are not added as though they were distinct original files.
What happened to the variants rejected by the network
The 20 failed publication records were Trial Reel variants of 16 originals. The records retained the network's rejection. A subsequent check reproduced an account eligibility restriction for that format.
On October 26, generation of new Trial Reels was disabled for ZaTek AI. Normal publishing settings were preserved. The rejection was not recorded as a successful publication and the variants were not republished to prepare this case.
How each result was counted
- Job
A processing identifier created within the period. It can represent an original, a variant or reuse.
- Source file
A distinct SHA-256 fingerprint across completed jobs, also deduplicated between the two installations.
- Confirmed publication
A receipt linked to an eligible job, with completed state, a network identifier, link and date within the period. Counted once per installation, network and identifier.
- Excluded tests
Explicit test or QA flags, source fingerprints and identified derivatives were checked. A missing flag does not establish that all remaining historical records were free of testing.
What this review cannot calculate
The records used do not contain a comparable manual working-time baseline or a complete cost per job for this period. Duration data only covers part of the material and does not represent a general processing time.
The case therefore reports states, files and receipts. It does not present hours saved, ROI, reach, impressions or sales as measured results. Publication figures refer to historical receipts within the period.
What this case contributes to another automation
The workflow shows decisions that also matter when connecting documents, CRMs or other systems: defining when each step is complete, preserving pending work and distinguishing a tool's response from a confirmed outcome.
- Verifiable states
Separate a received input, a prepared job, review and a confirmed output.
- Controlled recovery
Check what happened before repeating an action and keep rejections visible.
- A specific scope
Choose a workflow and agree on the records that allow the operating team to accept it.
Aggregate data and methodology
The CSV contains the counts for each installation and the total. The methodology explains the period, units and limits of the review.