Two active Performance Max campaigns, both feed-only (Shopping placements, no text/image/video assets, all automations opted out). On Triple Whale numbers, both are performing well above their targets — the question is how to scale them safely and what to build next.
| Campaign | Segment | Budget/day | Target ROAS | Spend 14d (pace) | Triple Whale revenue | Orders | ROAS | CPA |
|---|---|---|---|---|---|---|---|---|
| LD_GLP_PDP_US_PMAX_GENERIC_tROAS | Non-Brand | $2,000 | 2.25 | $19,129 · 68% of budget | $84,240 | 1,020 | 4.40 | $18.75 |
| LD_GLP_PDP_US_PMAX_BRAND_tROAS | Brand | $500 | 5.50 | $5,477 · 78% of budget | $61,605 | 809 | 11.25 | $6.77 |
May 28 – Jun 10 · revenue/orders = Triple Whale last-click; spend & budgets = Google Ads.
Neither campaign spends its full budget (68% and 78%), so raising budgets alone won't add volume — the target ROAS settings are the brake. GENERIC's platform-visible ROAS (~2.0–2.4) sits at its 2.25 target, but Triple Whale measures it at 4.40 — the campaign has roughly 2× more real headroom than the bidding engine believes. The safe scaling path: fix conversion capture first (see Search page, same fix benefits both accounts), then walk GENERIC's target down in 0.15 steps (2.25 → 2.10 → 1.95...) one week at a time, watching Triple Whale ROAS ≥ 3.0 as the floor. Same playbook for BRAND with a 4.5 Triple Whale floor.
The GENERIC campaign keeps brand traffic out using a shared negative-keyword list containing exactly one entry: "ledisa" (phrase match). Google's purpose-built brand exclusions for PMax (brand lists) are not set. The negative works — generic search-term categories show no "ledisa" variants — but it has no coverage for misspellings: "leidsa" got 671 clicks in 30 days and isn't blocked anywhere.
Identical product IDs in both listing groups — separation is purely query-based. The BRAND campaign (target 5.5) also captured ≥1,189 visible clicks on generic categories ("glp 1 patches", "weight loss patch") in 30 days — generic traffic winning at brand-level economics. Fine while it converts at 11.25, but quantify it before any target change.
(1) Add misspellings ("leidsa", "ladisa", "ledissa") to the shared negative list. (2) Configure the official brand exclusion (brand list: Ledisa) on GENERIC as the structural safety net. (3) Re-check the BRAND campaign's generic share monthly via search-term insights.
The feed is 20 keyword-permuted listings of the same GLP-1 patch (titles rotate "wellness and balance", "focus and consistency", etc.). Top 10 take 94% of product spend. Zero waste: no product spent more than $50 without converting. Product-level revenue is only available platform-side (Triple Whale tracks per campaign/ad group, not per feed item) — so the table below is Google-reported and should be read for relative comparison, not absolute return.
| Item | Listing title | Spend 90d | Conv. | Relative ROAS* |
|---|---|---|---|---|
| glp1_patch_mv15 | "Fastest Way to Lose Weight – GLP-1 Weight Loss Patch" | $36,088 | 1,413 | 2.34 |
| glp1_patch_mv12 | "GLP-1 Patches – Premium Daily Weight Loss Support" | $27,823 | 1,631 | 3.45 |
| glp1_patch_mv11 | "Weight Loss Patch for Women" | $7,614 | 670 | 5.32 |
| *Google-reported, for ranking products against each other. The clear pattern: mv15 ("Fastest Way to Lose Weight") takes the most budget with the weakest return AND the most aggressive title — the exact title style a policy sweep would hit. mv12/mv11 earn 1.5–2.3× more per dollar with softer titles. | ||||
The entire $54K/month rides one Merchant Center account whose titles lead with "Fastest Way to Lose Weight". One feed disapproval sweep = both campaigns dark. Recommendation: prepare a compliant-title supplemental feed as insurance, and shift listing-group weight toward the softer-titled winners (mv12/mv11 style).
Berberine, NAD+, Sleep and Energy — heroes on ledisa.com — have essentially no Shopping presence. A Berberine listing test rides the same renaming wave the category leaders already took (see Competitors) and opens a cheaper keyword universe (see Search page, advertorial research).
Today a single product feed powers both the BRAND and GENERIC campaigns, with separation enforced only by that one negative keyword. A cleaner, more durable structure is two purpose-built feeds, each optimized for the job its campaign does.
The feed behind the BRAND campaign should put the brand name at the very front of every title — "Ledisa GLP-1 Patch — Daily Weight-Loss Support" rather than "Fastest Way to Lose Weight – GLP-1 Patch". When someone searches "ledisa patches", Google matches and bolds the leading brand token, lifting relevance, click-through and Ad Rank on exactly the queries this campaign exists to win — and it reinforces brand recognition for everyone who scrolls past.
The feed behind the GENERIC campaign should lead with the category and benefit, brand later — "GLP-1 Weight-Loss Patch for Women — Berberine Formula | Ledisa". Non-brand shoppers search by problem and ingredient, not by brand, so the matching tokens belong up front. This is also where the compliant-title rewrite lives (softening the "Fastest Way to Lose Weight" style that invites policy review).
One feed forces a single title to serve two opposite intents and compromises both. Two feeds let brand titles maximize brand relevance and non-brand titles maximize category relevance — and give a structural brand/non-brand boundary at the feed level, on top of the negative-keyword and brand-exclusion boundaries, so brand and generic traffic stop competing in the same auctions.
Brand feed: Brand → Product → Key Benefit → Format/Count — e.g.
"Ledisa GLP-1 Patch · Appetite & Weight Support · 30 Patches".
Non-brand feed: Category → Audience/Benefit → Ingredient → Brand — e.g.
"GLP-1 Weight-Loss Patch · for Women · Berberine · by Ledisa". Both stay within structure/function language;
no drug or "fastest/guaranteed" claims.
Every Performance Max campaign this account has ever run has been feed-only with zero creative assets — so an asset-based PMax has never been tried. There is no past failure to worry about and no past success to assume; it is a genuine, clean test.
4 weeks minimum. Success = Triple Whale ROAS ≥ 3.0 at ≥$150/day spend. Kill rule = Triple Whale ROAS < 2.0 after 4 full weeks. All evaluation on Triple Whale, not platform numbers.
Brand demand currently flows through PMax BRAND at $6.77 CPA. A standard Shopping campaign dedicated to brand queries gives direct control over bids and query shaping that PMax can't offer — useful both as a scaling lever and as insurance.
Compare against PMax BRAND on Triple Whale CPA/ROAS after 3–4 weeks. If the Shopping campaign matches ≤$8 CPA with better query control, shift brand budget toward it; if not, keep PMax BRAND and retire the test.