46/100 reads Critical, but this is not a keyword problem: 94% of flagged unprofitable spend sits on the core drug terms (viagra, sildenafil, tadalafil, cialis) that are the front door of the business — converting, just 10–25% above the $215 break-even you are deliberately exceeding during tROAS recovery. Nothing needs pausing. The real fixes are ad copy/offer (Expected CTR is Below Average on 373 core keywords per the fresh QS audit), the brand LP, and $0-cost hygiene: 582 zombies and 387 duplicate terms.
Blocking — resolve these before touching keywords
Check Metabase (your source of truth, not Google) that core terms are performing to the recovery plan. Confirms the flags reflect the deliberate above-break-even posture during volume rebuild, not a real efficiency loss — and unblocks a confident 'leave them running'.
Blocks all keyword-level action until confirmed
The fix lives upstream (ads, LP, targets) — not in keyword edits
Expected CTR is Below Average on 373 core drug-term keywords (fresh Quality Score audit, 2026-07-08, 43/100) — ads are keyword-relevant but not compelling vs Hims, Ro and BlueChew. That is the CPA-inflation mechanism behind the unprofitable flags. Recovering Ad Rank lowers CPCs and brings core terms back under break-even. Run once tROAS recovery stabilises (mid-July).
Recover Ad Rank → lower CPCs → core terms back under $215 break-even (~$59k affected spend) · /rsa-maker + /offer-auditor
| Keyword | Campaign | Cost | Conv. | CPA | QS | Note |
|---|---|---|---|---|---|---|
| viagraEXACTcorewatch | ED - All - High | $8,884 | 37.8 | $235 / $215 | 7 | |
| sildenafil buyEXACTcorewatch | ED - All - Low | $6,059 | 27.9 | $217 / $215 | 7 | |
| viagra pillEXACTcorewatch | ED - All - High | $4,024 | 12.6 | $319 / $215 | 7 | |
| tadalafilEXACTcorewatch | ED - All - High | $3,360 | 13.8 | $243 / $215 | 5 | |
| generic cialisEXACTcorewatch | ED - All - High | $2,574 | 9.3 | $276 / $215 | 7 |
The fresh QS audit flags "promo code for ziphealth" at QS 2 and "zip health reviews" at QS 0 — a cheap Ad Rank win on high-value brand searches.
Cheap Ad Rank win on high-value brand searches · /lp-auditor
65% of flagged unprofitable spend ($41.0k) concentrates in this one campaign — the campaign target, not individual keywords, is the right lever. Validate the July re-tighten to 93% tROAS balances position vs efficiency during recovery.
Right target for position vs efficiency during recovery ($41k of flagged spend) · /bidding-auditor
Try these before any pause is considered
These keywords are converting above break-even — recover, do not remove. In order: (1) /search-term-auditor ngrams over 120–180d to isolate bad n-grams to exclude instead of pausing (existing report 2026-05-27 is stale — re-run); (2) /offer-auditor — higher CVR lifts every keyword in the campaign (2026-05-19, stale); (3) /lp-auditor (2026-06-03, stale); (4) /strategy-specialist to validate whether tROAS targets are realistic during recovery.
Recovers ~$59k of core-term spend without losing front-door traffic · /search-term-auditor ngrams
Safe regardless of the open questions above
Zero impressions over the period, $0 cost — pure signal hygiene, safe regardless of the cascade. Examples below; the optimizer removes the full set with a dry-run first.
Signal/hygiene cleanup — $0 spend affected · /keyword-optimize cleanup
| Keyword | Campaign | Cost | Conv. | CPA | QS |
|---|---|---|---|---|---|
| ed meds onlineBROADcore | ED - All - NonComp | $0 | 0 | — | 8 |
| most effective ed medicationEXACTcore | ED - All - NonComp | $0 | 0 | — | — |
| cheap ed medicineEXACTcore | ED - All - NonComp | $0 | 0 | — | — |
| ed dysfunction medicationsEXACTcore | ED - All - NonComp | $0 | 0 | — | — |
| buy ed medicationEXACTcore | ED - All - NonComp | $0 | 0 | — | — |
Same keyword living in multiple campaigns (mostly brand terms sitting in Brand + ED campaigns on the same criterion) — self-competition and muddied signal. Safe structural fix.
Reduce self-competition · /keyword-optimize duplicates
| Keyword | Campaign | Cost | Conv. | CPA | QS |
|---|---|---|---|---|---|
| cialisEXACTmedium | ED - All - NonComp ↔ ED - All - NonComp | $272 | 0 | — | — |
| cialisEXACTmedium | ED - All - NonComp ↔ ED - All - High | $4,102 | 25.4 | — | — |
| cialisEXACTmedium | ED - All - NonComp ↔ ED - All - Low | $760 | 3 | — | — |
| generic viagra onlineEXACTmedium | ED - All - NonComp ↔ ED - All - Low | $1,202 | 4 | — | — |
| viagraEXACTmedium | ED - All - NonComp ↔ ED - All - NonComp | $545 | 4.4 | — | — |
Brand terms (ziphealth viagra, ziphealth tadalafilo, tadalafilo ziphealth) are serving in ED campaigns — $156 spend, 0 conversions — while a dedicated Brand campaign exists. Add negatives so brand searches route to Brand.
$156 rerouted to the Brand campaign · /keyword-optimize match-type
424 cross-campaign overlaps minus the 295 intentional ED - All - NonComp carve-out rows leaves ~129 genuine overlaps causing auction overlap between High and Low.
Reduce auction overlap · /keyword-optimize match-type
Profitable keywords worth a bid/scale decision
OVER_TARGET means profitable by the break-even gate but above the campaign's own target — a bid question, not a pause candidate. Both ED rows are actually under the $215 break-even. Also here: hims (conquest term, $2.1k / 8.1 conv / $263 CPA) — consistent with business.md's selective-conquest strategy, so a bid/target decision rather than a pause.
Scale/bid decisions on profitable terms
| Keyword | Campaign | Cost | Conv. | CPA | QS | Note |
|---|---|---|---|---|---|---|
| generic cialis onlineBROADcorewatch | ED - All - Low | $213 | 1 | $213 / $215 | 8 | Under break-even — leave running; bid review only |
| pastillas para el sexoEXACTwatch | ED - All - Low | $211 | 1 | $211 / $215 | 4 | Under break-even — leave running; bid review only |
| himsEXACTwatch | ED - All - Low | $2,140 | 8.1 | $263 / $215 | 5 | Converting conquest term above break-even — bid/target question per business.md selective-conquest strategy |
| himsBROADmaterial | ED - All - High | $679 | 2 | $340 / $215 | — | Converting conquest term above break-even — bid/target question per business.md selective-conquest strategy |
Full narrative: context/analysis/review-queue.json is generated from context/analysis/keyword-audit.md — open it from the Reports list or ask in chat. Mutations always go through /keyword-optimizer with a dry-run and your approval.