Content Insights

Content Insights (Content → Insights, /dashboard/content/insights) analyzes the videos posted about your shop to tell you what kind of content actually works: which products convert viewers into buyers, how long videos keep earning views, which hooks and selling points outperform, which creators are surging, your best individual videos, and which creative formats lift performance.

Much of this page is powered by AI video analysis: Cruva's AI watches and transcribes your videos, extracting hooks, selling points, tone, integration type, face visibility, lighting quality, and more.

Plan availability: the entire page requires the Growth plan or higher; lower plans see an "Upgrade Now" card.

A date range picker at the top right (default: the last 31 days) scopes every section. A sticky navigation bar jumps between the seven sections: Product Conversion, Video Lifespan, Hooks, Selling Points, Fast Risers, What's Working, Content Format.

Key metrics used throughout this page

• GPM (Gross Per Mille) — GMV per 1,000 views. The standard creator-economy efficiency metric.
• Lift — GPM for a bucket (a hook, selling point, format value, or single video) divided by your shop's baseline GPM. A 2.0× lift means it earned double the GMV per view of a typical video on your shop. Lift is sample-size blind — a 3× lift on 5 views prints the same number as 3× on 5 million views — so use Impact for prioritization.
• Impact (incremental GMV) — dollars earned above (or below) what the shop baseline GPM would have produced on the same views. Positive = genuine contribution; negative = underperformer. Impact is volume-aware: small buckets can't produce big impact by definition.

In lift tables, a small evidence dot next to each label is sized by view volume — bigger/darker dots mean the number is backed by more evidence. Lift values render green at ≥1.1× (bold ≥1.5×) and red below 0.7×.

1. Product Conversion Rate

Which products convert best per video view, in the selected date range.

A searchable, sortable table (10 rows per page, default sorted by Video Conv % descending) with one row per product:

• Product — name and image. Products with under 1,000 views get a "Low volume" warning badge because their conversion percentages may not be reliable.
• Videos — number of videos featuring this product.
• Impressions — how many times the product appeared in the TikTok Shop tab.
• Views — video views across all videos featuring this product.
• Avg Views / Video — Views ÷ Videos.
• Units — units sold.
• View-through % — Impressions ÷ Views: how often a video view led to the product page being shown.
• Sell-through % — Units ÷ Impressions: the share of shop-tab impressions that led to a sale.
• Video Conv % — Units ÷ Views: the share of video views that led to a sale. The headline conversion metric.
• GMV / 1K Views — revenue earned per 1,000 video views for this product.
• Total GMV

Export CSV downloads the full result set (honoring the current search and sort), not just the visible page.

Use this to pick which products to push in briefs and sampling: a product with high Video Conv % and GMV/1K Views deserves more creator content.

2. Video Lifespan

How long your videos keep earning meaningful views. Lifespan is defined as the days from posting until a video's daily views fall below 10% of its peak.

Only videos posted within the selected range that are at least 21 days old are analyzed, and videos still above the threshold (still actively earning views) are excluded from the lifespan calculation.

Left-side stat cards:

• Median lifespan (in days) — the headline number.
• Analyzed — total mature videos considered.
• Declined — videos that have decayed below the 10%-of-peak threshold (these define the distribution).
• Still active — videos still earning above the threshold.

The chart is a histogram of lifespan buckets with the percentage of videos in each bucket labeled on top.

Use this to time your content cadence: if the median lifespan is ~9 days, you need fresh videos at least that often to keep momentum.

3. Hooks (Hook Effectiveness)

Hook phrases — extracted by AI from the first few seconds of each video — ranked by effectiveness.

Controls:

• Product filter — restrict to videos featuring one product.
• All / Reoccurring toggle — "Reoccurring" keeps only hooks used in at least 2 videos, filtering out one-off phrasings.
• Export CSV — full list with all metrics.

Table columns (10 rows per page, default sorted by Impact descending; all sortable):

• Hook — the phrase, with the evidence dot.
• Impact — incremental GMV vs. shop baseline (see definitions above).
• Lift — bucket GPM ÷ baseline GPM.
• Videos — number of analyzed videos containing this hook.
• GMV — total GMV from videos in this bucket during the period.
• Views — total views in the bucket.
• GPM — GMV per 1,000 views for the bucket.
• Play button — opens the top-GMV video using this hook. When multiple videos use the hook, a badge shows the count and clicking opens a picker listing every video (by @handle and views). Clicking anywhere on the row does the same.

Use the winning hooks verbatim in creator briefs — they're proven openings for your products and audience.

4. Selling Points (Selling Point Effectiveness)

Identical table and controls to Hooks, but each row is a product claim or selling message extracted by AI from each video (e.g. "waterproof", "cheaper than X", "one-handed opening"). Same columns (Selling Point, Impact, Lift, Videos, GMV, Views, GPM, play button), same All/Reoccurring toggle, product filter, and CSV export.

Use it to learn which claims actually move product, and put those front-and-center in briefs and product descriptions.

5. Fast Risers

Creators whose total views in the selected date range outpaced their views in the previous period (the equally-long window immediately before the range) — your currently-surging creators.

To qualify, a creator's videos need at least 100 views in the previous period (this prevents divide-by-tiny-number artifacts). If the table is empty, try widening the date range or picking a more recent window.

Columns (10 per page):

• Creator — @handle plus how many videos they posted in range.
• Views — total views in the range.
• GMV — GMV in the range.
• Growth × — views this period ÷ views previous period. Hovering shows the previous period's raw view count.

Export CSV downloads the full list.

Fast risers are prime candidates for immediate follow-up: more samples, higher commission, or a targeted-collab offer while their momentum is hot.

6. What's Working (Top Performing Videos)

A card grid of your best-performing videos in the selected date range. The Sort by toggle switches between four definitions of "best":

• Impact (default) — a combined score weighing both how much GMV the video drove and how much better it performed than your typical video (evidence-weighted dollars above baseline).
• Lift — how much better the video converted views into GMV vs. the shop baseline.
• GMV — raw revenue generated in the range.
• GPM — GMV per 1,000 views (revenue density).

Each card shows the video thumbnail (hover reveals a play button; clicking opens the full video modal), an AI-detected format badge (integration type or tone — used to group similar formats in the Content Format playbook), the creator's @handle, and four metric pills: Impact ($, signed), Lift (×), GMV, and GPM. Six cards per page.

Export CSV includes every video with creator, title, post date, GMV, views, likes, comments, units, GPM, engagement %, conversion %, lift, and impact score.

7. Content Format (Format Playbook)

Lift analysis by creative format dimension, all AI-detected from the videos themselves. Buckets need at least 2 videos to be included. The five dimensions:

• Tone — the emotional register of the video (casual, energetic, informative, …).
• Integration Type — how the product appears (demo, unboxing, lifestyle, review, …).
• Face Visible — whether the creator's face is on camera.
• Product in First 3s — whether the product is visible in the first 3 seconds.
• Lighting Quality — overall lighting quality as rated by AI.

Combination cards

• Most Successful Combination — the highest-lift value in each dimension, with the per-value lift. "Combined lift" is the geometric mean of the per-dimension lifts — a rough estimate of the compounded effect if a video hit all of them.
• Most Common Combination — the most frequently used value in each dimension (with video counts): what creators are actually making today.
• Most Viewed Combination — the value in each dimension with the most total views: what pulls attention regardless of whether it converts.

Comparing "Most Successful" to "Most Common" is the point: when they differ, there's an untapped format your creators aren't using enough — put the winning combination in your briefs.

Dimension cards

Below the combinations, one card per dimension shows the top value (with its lift) and horizontal lift bars for up to four values — green bars for lift ≥ 1×, red for below.