The cull
22 presells demoted, not patched
Strict SV-baseline audit demoted 22 presell pages from paid to SEO-only. The remaining 3–4 paid-active pages all anchor on the DETOX hero SKU - the page count went down on purpose, not as failure.
A new 'vitamin diffuser' category fighting a Google 'vape' keyword ban (every vape keyword prices at $0 CPC), an over-built 25-page presell network with insufficient SV baseline, and 5 dead products plus 7 losing variants burning ~32% of spend.
The strategic frame every angle traces back to - the villain we name, the mechanism that beats it, and the proof no competitor can copy.
Google 'vape' keyword ban (every vape KW = $0 CPC) + over-built presell sprawl + 5 dead products burning ~32% of spend
Demote-22-to-SEO + consolidate to a single-SKU DETOX PMax test + GMC compliance hard-set
Feed positioned in Vitamins & Supplements (not generic Health & Beauty) + 98/100 Merchant Center quality score + diffuser-only terminology
Strict SV-baseline audit demoted 22 presell pages from paid to SEO-only. The remaining 3–4 paid-active pages all anchor on the DETOX hero SKU - the page count went down on purpose, not as failure.
Three Shopping campaigns consolidated into one PMax test focused entirely on the DETOX hero SKU - 20 product images, 28 search themes, full creative coverage across headlines, long headlines, descriptions and landscape / square / portrait imagery.
Single SKU, single PMax bet, 21-day learning runway.
10 zombie PMax-Presell campaigns paused, 10 already-removed campaigns force-deleted. GMC re-segmented into BOFU / MOFU / TOFU at 16/14/7 products with 87 unit-excluded items. Account state goes from sprawl to focus.
Feed category set to Vitamins & Supplements, not the generic Health & Beauty default. Copy uses 'diffuser', 'inhaler' or 'pen' exclusively - never 'vape'. Every vape keyword prices at $0 CPC, so the policy ban is enforced upstream of bidding.
Single DETOX variant pulled from the 90-day window before consolidation - $649 spend returned 1.36× direct ROAS and 7.86× attributed-funnel. The strongest variant in the line hit 21.7× funnel ROAS - cited as the bet justification for the focused PMax consolidation, not as an account-level number.
Feed audit (2026-05-07) showed 5 dead products + 7 losing variants accounting for ~32% of spend. 29-item product exclusion deployed to PMax-Feed BOFU / MOFU / TOFU listing groups.
37 active products at 100% approval. Three consolidated segments (BOFU 16, MOFU 14, TOFU 7) with 87 unit-excluded items - feed integrity preserved through aggressive consolidation.