The clip desk that reads its own drop-off
A 60-to-99-person full-service agency's social lane had 2 editors at capacity cutting a weekly video podcast into clips, and its AI clipping tool was acquired and shut down.
Clip picking and recuts rely on editor gut calls
A fixed-rule desk scores each podcast transcript and proposes 10 clip candidates with a hook line and reason. Forty-eight hours after posting, it reads client-exported retention files, pins each viewer drop to the transcript line playing, and proposes recuts that only remove lines. Editors approve, edit or reject each one with a written reason; the agency owns the code.
- the desk pinned 6 of 6 drops to a transcript line and an export row; against the drops we designed into the curves ourselves, it found 6 of 6 and 0 of 1 decoy dips fired
- all 5 recut proposals ended the week with a logged editor decision and a reason (100%, by construction); 1 decision bounced for having no reason
- Rule-based transcript scorer
- Retention curve analysis
- Client exports
- Official platform APIs
- Editor approval queue
- Rules file
Same problem?Send it over.
Tell me what your version looks like and which tools it runs in. I reply with how I would build it.
-
Performance marketing
The retainer, on the record
all 6 approved spend changes carry a named approver, and the gate sent 1 approval back for a missing reason
-
Media and creative agency
The Monday margin watch that reads jobs, not people
1,124 timesheet rows read, 31 person ids dropped at load, 0 found in any output file
-
Brand agency
The coordinator job we rebuilt before it was posted
7 of the 9 bullets on the drafted coordinator posting now run as fixed workflows; 2 stay with people