🎯 This week's prioritized changes — specific creative × ad set actions, ranked by ₹ impact
Top creatives · spend vs link-click CTR
Each row = one creative aggregated across all the ad sets it runs in (Meta makes a separate ad-entity per deployment; we roll them up so the same creative isn't repeated 7×). Bars = total spend. Dashed line = weighted link-click CTR. Compare to the 1% benchmark dotted line.
Creative funnel · impression → hook → hold → completion
Industry benchmarks shown as light grey bars (Hook 25% · Hold 35% · Completion 5%). Anything above = healthy; below = address.
🧬 Creative performance · aggregated across all ad set deployments
· weighted by impressions
Each row groups all deployments of the same creative across different ad sets. Use this to decide which creatives to scale, maintain, refresh, or kill.
Creative
# Deployments
Total Spend
Weighted Link CTR
Total Purchases
Blended CPA
Verdict
🪝 Hook vs Hold diagnostic — identifies where weak creatives fail
Hook = 3-sec views ÷ impressions (do they stop scrolling). Hold = thruplay ÷ 3-sec views (do they stay). A low Hook means the first 1.5s isn't grabbing them; a low Hold means the body falls flat. Benchmark: Hook ≥ 25%, Hold ≥ 35%.
Creative
Spend
Hook %
Hold %
Where it fails
Audience efficiency map · reach vs CPM (bubble = spend)
What this shows: Each bubble is one ad set. X-axis = audience reach; Y-axis = CPM (lower is cheaper). Bubble size = spend. Read it: bottom-right is ideal (large reach + low CPM); top-left is wasteful (small reach + expensive). Move budget from top-left to bottom-right.
Frequency distribution
X-axis: frequency bucket (avg times each user saw an ad in this ad set). Y-axis: number of ad sets in that bucket. Read it: bars to the right of 2.5 are fatigued and need creative rotation or audience expansion.
⚠ Pairwise audience-overlap risk · top 5 ad-set pairs
— estimated, based on naming similarity + interest tokens
Top 15 ad sets · ranked by spend
Action
Ad Set
Spend
CPA
CPM
Freq
Reach
Exhaustion
Pareto · cumulative spend vs cumulative purchases
What this shows: Ad sets sorted by spend (largest first). Blue line = cumulative % of spend; Orange line = cumulative % of purchases. Read it: if 50% of spend produces only 20% of purchases (orange far below blue), money is misallocated — shift budget toward ad sets where the orange line catches up to blue faster. Ideal: orange tracks ≥ blue at every point.
Efficiency index per ad set · spend-share ÷ purchase-share
What this shows: Each bar = one ad set. Index = (% of total spend) ÷ (% of total purchases). Read it: index < 1.0 (green) means the ad set delivers MORE purchases than its budget share — scale it. Index > 1.5 (red) means it's burning budget without commensurate purchases — cut or restructure. Index near 1.0 = pulling its weight.
Top 15 ad sets · ranked by spend
Action
Ad Set / Event
Spend
Budget
Purchases
CPA
Eff Index
Spend Share
🧬 Creative knowledge base — every creative that ran in the last 12 months, ranked by Creative Quality Score (CQS)
CQS = a 0–100 score that blends mature link-CTR (30%), VR@25% (15%), VR@50% (15%), hook rate (10%), hold rate (10%), and spend-attribution proxy (20%). Each component is a percentile rank within your account history. ≥70 = winner, 50–69 = solid, <50 = needs work.
Filter by tags below. Click a row to see its component breakdown.
Filter:
Creative
Tags
CQS
Link CTR
Hook
Hold
VR@25
VR@50
Spend
Impressions
Ad sets
📊 Tag-segment performance — which tags consistently produce winners?
For each tag value, shows median CQS of creatives carrying it (impression-weighted). Use this to see e.g. which talents or topics correlate with high scores.
🤖 AI Creative Scorer
Drop a video. We extract 3 frames (hook, middle, CTA), send them to Claude vision, get back tag classifications across 18 dimensions, then score against your 33-creative knowledge base. The whole pipeline runs in your browser using the Cowork Claude bridge — no manual tagging.
📹
Drop your new creative here
or click to browse · MP4 / MOV / WebM / JPG / PNG
🔴 Fatigued deployments — both triggers fired (T1 + T2)
All flagged deployments are now paused — no creative is currently burning budget while fatigued. Listed below for retrospective context (and so the deployment is not re-launched without a refresh).
Creative
Ad Set
Status
Spend
Last 3d Freq
T2 Severity
T1 Middle CTR
T1 Final CTR
Drop
🟡 Monitor list — one trigger fired; not refresh-urgent yet
Type
Creative
Ad Set
Spend
T2 Severity
T1 Middle CTR
T1 Final CTR
Drop
🔍 Same creative, different verdicts — why deployment granularity matters
📖 How fatigue is detected
An ad is evaluated for fatigue if it meets all three: ≥₹30,000 cumulative spend, ≥30,000 cumulative impressions, and ≥3 days with impressions > 100, over the 60-day window Apr 3 – Jun 1, 2026. CTR throughout is link-click CTR (link clicks ÷ impressions), not Meta's all-CTR.
Two triggers, four states:
Trigger 2 (peak-vs-current): 3-day rolling link-CTR has dropped ≥25% from its observed peak. Peak window must have ≥8,000 impressions each day (filters early-day spike noise — raised from 1,000 to suppress small-sample false peaks).
Trigger 1 (sustained-decline): compares the ad's middle 50% of impressions (skipping first 25% and last 25%) to the final 25%. Confirms if final ≤ middle × 0.80.
🔴 FATIGUED-CONFIRMED: Both triggers fire. High confidence — refresh now.
🟡 WATCH (noise risk): Only T2 fires. Likely an early-day CTR spike inflated the peak; the calm middle is stable. Monitor but don't refresh.
🟡 WATCH (slow drift): Only T1 fires. No sharp drop yet, but middle→final is sliding. Plan refresh in next ~5 days.
🟢 HEALTHY: Neither fires.
Granularity: each row is a unique creative × ad set deployment. The same creative running in two ad sets is treated as two separate evaluations because audience saturation differs per ad set.