How RaidCut uses performance data

Why import YouTube Studio and TikTok exports

RaidCut can score and package clips before they go out, but it can only get smarter if it sees what actually happened after you posted. Importing your platform exports closes the loop: RaidCut compares the clips you approved against the views, engagement, and timing they earned, so its future suggestions are grounded in your channel's reality instead of generic best-practices.

What data RaidCut looks at

• Views — the primary signal of reach for each clip.
• Timing — the date and time a post went live, used to learn your best windows.
• Title — the published title text, compared against what performed well.
• Platform — YouTube vs TikTok behave differently and are learned separately.
• Clip identity — which RaidCut clip a row maps to (via auto-link or manual link).
• Performance period — the date range the export covers.

Why manual linking of unmatched videos matters

Platform exports identify each video by the platform's own ID. RaidCut auto-links what it can, but some rows stay unmatched. Until a row is linked to a clip, its views can't teach RaidCut anything. Spending a minute linking your top unmatched videos — sorted by views in the app's Performance page — has the biggest impact, because your best performers carry the most signal.

How performance data feeds better titles and descriptions

Once clips are linked to real numbers, RaidCut can tell which title formats and phrasings tend to do well for you and which underperform. That feedback flows into the title and description suggestions you see during review, while still respecting your brand-voice rules and guardrails.

How performance data helps posting cadence

Timing data lets RaidCut propose posting windows that have worked for your audience — for example, a particular weekday and hour on a given platform. These show up as scheduling suggestions, never as forced changes.

Why performance data can become stale

Channels change. A title pattern that worked last month may fade, and new content shifts your best windows. If your most recent import is older than your freshness setting, RaidCut shows a Performance data stale badge and treats its guidance more cautiously. Import a fresh export to bring learning back up to date.

Privacy and trust

Performance data is imported locally and used only to improve your own clips and schedule. It is not exposed publicly, and RaidCut does not turn your analytics into something it shares. The goal is a better workflow for you, not a data product.

A note on low confidence

Learning is only useful once enough clips have real performance data behind them. With a small sample, treat suggestions as low-confidence hints rather than rules. The more clips you link to real results, the more dependable RaidCut's guidance becomes. See how learning snapshots work.