Google Ads Audit — Grow Fragrance

Severity 1 — Fix this week

1.1 Branded-Search has 732 negative keywords stuffed in, blocking brand-intent queries.

• The "Conflicting negative keywords" warning lists [grow fragrance discount code] as a blocked keyword. Users searching this are at the bottom of the funnel.
• The 732 negatives are competitor terms (Aera, Aesop, Abbey & Sullivan, Aerie, Aery Living) and category terms ("100 natural air freshener", "soy candles", "fragrance company") — these belong on a non-brand campaign, not the brand campaign.
• Action: Review the negative list, keep only negatives that are genuinely competitive-confusion (e.g., "grow taller" type misintents). Remove all category/competitor negatives — they make no sense on a brand campaign.

1.2 Branded-Search is "Limited by budget" at $280/day.

• Brand search should never be capped. Every brand-query click is recapturing existing demand at the lowest CPC available.
• Google's own recommendation: raise to $480/day (estimated +28.6 conversions).
• Action: Either uncap brand entirely or set a budget high enough that "Limited by budget" never triggers. The risk of overspending on pure brand terms is structurally low because brand query volume is finite.

1.3 Brand ROAS is 6.59x at a 5.14x target — suspiciously low for pure brand.

• Healthy DTC brand search typically returns 8–15x ROAS because the buyers are pre-existing intenders.
• 6.59x on brand suggests either (a) the campaign is bleeding into non-brand via the paused "Branded-Keywords-Broad" historical setup, (b) attribution misallocation where PMax-generated demand is getting credited to brand search, or (c) ad copy/landing pages aren't converting brand traffic efficiently.
• Action: Pull a search-terms report on Branded-Search for the last 90 days. Confirm only brand queries are triggering. If non-brand is leaking in, that's a structural fix.

1.4 PMax campaigns have no audience signals.

• Every asset group on Scents-PMAX-2026 says "No audience signals provided."
• PMax uses audience signals as a seed for its targeting algorithm. Without them, the algorithm spends 2–3x longer in the learning phase and often finds suboptimal audiences.
• Action: Add audience signals to each asset group. At minimum: your converters in the last 90 days, your AOV-above-median customers, your top-fragrance-category buyers. Klaviyo + GA4 + Shopify data via Windsor can build these segments.

Severity 2 — Fix this month

2.1 Upper-funnel conversion actions are misconfigured.

• Add to cart, Begin checkout, Page view, and Download all show "Misconfigured" status in Tools > Goals > Summary.
• Smart bidding (PMax + Branded-Search) feeds on these upper-funnel signals to learn faster and find users earlier in the journey. Without them, the algorithm only has Purchase events — sparse and slow.
• Action: Re-fire the Google Ads tag for these events. Verify in Tag Assistant. Mark them as "Secondary" conversions (counted but not bid-target).

2.2 PMax asset groups all show "Average" ad strength.

• Every active asset group (Woodland-Sage, Golden-Grove, Pacific-Driftwood, Midnight-Orchid) is at "Average." Target is "Excellent" — Google de-prioritizes Average groups in auction.
• Action: Per asset group, add 5+ more headlines, 4+ more descriptions, 5+ more images of varying sizes, and at least 1 more video. Aim for the "Excellent" badge.

2.3 Midnight-Orchid asset group is unprofitable at 1.66x ROAS.

• Below the 2x floor needed to cover ~50% gross margin.
• Action: Pause Midnight-Orchid for a 2-week window. Confirm overall campaign ROAS holds, then either retire it or rebuild with better assets and a tighter listing group.

2.4 Algorithm concentrating 77% of Scents-PMAX spend on Woodland-Sage.

• Could be a real signal (sandalwood/sage genuinely outperform) or a learning local maximum.
• Action: Split into single-asset-group PMax campaigns per scent family. PMax mixes products poorly when asset groups vary widely in performance. Single-theme campaigns let you set budgets per scent and isolate performance.

Severity 3 — Monitor

3.1 Optimization score trending down (-2% over the period).

Most of the recommendations Google is pushing are upsells you should reject (Demand Gen new campaign, AI Max Beta, broad match keywords). The score will not recover until you either accept those or "dismiss with reason" — dismissing returns score, accepting bad recs hurts performance.

3.2 RSA auto-improvement is on.

The "Improve your responsive search ads" auto-apply is ON, which lets Google pull headlines/descriptions from your site into your ads automatically. Action: Turn this OFF in Recommendations > Auto-apply settings > Maintain your ads. Best practice is to control your ad copy.

3.3 No non-brand Search campaign.

All non-brand demand capture is happening implicitly via PMax search themes (sage scent candle, citrus scented room spray, etc.). PMax is fine for category capture, but a dedicated non-brand Search campaign with phrase/exact keywords gives you transparency and control over high-margin queries. Action: Consider a small non-brand Search test ($30–50/day) on your top 20 phrase-match category terms. Run for 60 days, compare ROAS to PMax category-theme performance.