Content moderation

Content Moderation (Content → Moderation, /dashboard/content/moderation) lets you scan every video posted about your shop for keywords — across the spoken audio, on-screen text, description, and hashtags — and flag the ones that break your rules. Typical uses:

• Disclosure compliance — flag sponsored videos that are missing #ad or #paidpartnership.
• Banned claims — flag videos whose transcript contains prohibited health/medical or exaggerated claims.
• Competitor mentions — flag videos that name competing brands or products.
• Brand-safety terms — flag any wording you don't want associated with your products.

The scanning works on Cruva's AI analysis of each video, so keywords are matched not just in the visible description/hashtags but also in the audio transcript (what the creator says) and on-screen text (text overlays in the video).

Plan availability: requires the Growth plan or higher. On lower plans the page is blurred with an "Upgrade Now" card. A help-video banner at the top of the page links to a walkthrough video.

The rules list

The main page is a table of your moderation rules:

• Rule — the rule's name.
• Last Scan — when it last ran ("Never" if it hasn't).
• Result — the outcome of the last scan: "n flagged" or "All passed".
• Actions — Run scan, Edit, View Details (scan history), and Delete.

Clicking a row opens the rule's detail page (scan history). New Rule in the header opens the rule builder. Deleting a rule asks for confirmation and permanently removes the rule and its scan history — this can't be undone.

Creating or editing a rule

Click New Rule (or Edit on an existing rule) to open the rule builder. The left side is the configuration form; the right side is a live preview of what the rule would flag.

Rule Name

A label for the rule, e.g. "Disclosure Check".

Scan (fields to check)

Choose which parts of each video to check for keywords — at least one is required:

• Audio transcript (selected by default) — everything spoken in the video, transcribed by AI.
• Video description — the caption text.
• On-screen text — text overlays detected in the video frames.
• Hashtags

Keywords

Two keyword lists, each holding up to 20 keywords. Type a keyword and press Enter (or the + button) to add it; keywords are stored lowercase. Click the × on a chip to remove it.

• Flag if present (include keywords) — flags a video when the selected scan fields contain any of these keywords. Use for competitor mentions, banned claims, prohibited terms.
• Flag if missing (exclude keywords) — flags a video that is missing all of these keywords. Listing several means any one of them counts as present. Use for required disclosures: e.g. list #ad and #paidpartnership, and a video is flagged only if it has neither.

A rule can use both lists at once. At least one keyword (in either list) is required to save.

Date Range (optional)

Limits which videos are scanned to those posted within the range. Leave blank for all time.

Live preview

As you edit, the right panel updates (debounced, so it doesn't refetch on every keystroke) with:

• A count line — "N videos would be flagged" (or "N+ videos would be flagged" when the count is capped).
• A sample table of up to 10 videos the rule would flag right now, with the same columns as scan results (thumbnail, creator/title, product, flagged tags, views, GMV, units, posted date). Clicking a preview row opens the video on TikTok.

This lets you tune keywords before saving — if the preview flags too much or too little, adjust and watch the count change.

Saving requires a name, at least one keyword, and at least one scan field. The builder warns you about unsaved changes if you try to leave mid-edit.

Running scans

Scans are run on demand: click the Run scan action on the rules list. When it finishes, a toast reports the result — "X of Y videos flagged." Scans evaluate the videos tied to your shop (limited by the rule's optional posted-date range).

There is no scheduler on this page — you re-run a rule whenever you want a fresh snapshot. Each run is stored in the rule's scan history.

Rule detail: scan history

Clicking a rule (or its View Details action) opens its scan-history page, listing the most recent runs (up to 10 kept per rule):

• Scanned — date/time the scan ran.
• Total — videos checked.
• Flagged / Passed — how many failed / passed the rule.
• Flag Rate — flagged ÷ total.

An Edit button in the header opens the rule builder. Clicking a scan row opens that scan's results.

Scan results

The scan detail page shows summary stats — Total videos, Flagged, Passed, Flag rate % — and a paginated table of the flagged videos:

• Video — thumbnail; clicking it (or the row) opens the full video modal, which also includes Cruva's AI analysis of the video (content overview, target audience, tone, product integration type, product screen time, face visibility, GMV, CTR, units sold, and more).
• Info — creator @handle and video title.
• Product — the product featured in the video (image and name), or "No product".
• Flagged Tags — exactly which keywords triggered the flag, shown as red badges.
• Views, GMV, Units, CTR — performance stats for the video.
• Posted — post date.

Two actions in the header:

• Export CSV — downloads the flagged results.
• Save as List — creates a new list containing the @handles of all flagged creators. Give it a name (e.g. "Flagged Creators — Jan 2026") and use it anywhere lists work in Cruva — for example as an outreach/automation audience to DM those creators about fixing their disclosures, or as a filter cohort in other reports.

Quick recipes

• Catch missing ad disclosure: Scan fields: Hashtags + Video description. Flag if missing: #ad, #paidpartnership, #sponsored. Run after each campaign wave, then Save as List and DM the flagged creators.
• Catch banned claims: Scan fields: Audio transcript + On-screen text. Flag if present: cures, guaranteed, fda approved, etc.
• Catch competitor mentions: Scan fields: Audio transcript + Video description + Hashtags. Flag if present: competitor brand names.