Where My Automation Stops: Judging Titles, Keywords And Covers By Hand

Description and use

Automation generates metadata, prevents double‑posts, and renders covers, yet manual review of titles, keywords, and cover readability remains essential, taking about 90 min per 40‑pack batch.

Problem

Automating metadata, uploads, and rendering for sticker packs fails when human judgment is required for titles, keywords, and cover quality.

Outcome

Automation handles metadata, double‑post prevention, and rendering, reducing manual effort to title, keyword, and cover review, which remains the most valuable 90 min per batch.

Workflow

• Run make_metadata.py to generate platform‑specific CSV headers
• Run upload_tracker.py to prevent double‑posting
• Run pack_product.py to render 1280×720 covers and 600×600 thumbnails
• Package assets into ZIPs with covers and metadata
• Manually review each CSV row for title and keyword quality
• Manually inspect each cover at thumbnail size for readability
• Select keeper images from AI generations
• Submit approved packs to stock platforms

Evidence

The script never fails on headers.

Source

I read every row. For 40 packs that's maybe 90 minutes.

Source

The ledger handles that cold.

Source

I look at every cover against the thumbnail size it will actually appear at.

Source