Lucerna.Skin

Creative slate · 2026-Q2 Meta launch
Prepared 2026-05-20 · Daniel Illingworth · Flowsly AI

Two campaigns, ten placement units. Prospecting locked on Carla’s first-batch references plus one Gemini-iterated variant. Retargeting half-locked: the reviews angle shipped today, three more angles wait on Carla’s second-batch revisions.

Status at a glance

Campaign A — LK_Prospect_2026Q2 · €8/day

Targeting — ad-set breakdown New 2026-05-20

Billy pushback 2026-05-20: at €8/day on a niche premium DTC product, broad targeting burns impressions on irrelevant audiences. Manual interest stacks required. Same 6 ads (A1–A6) run inside each ad set — Meta does not double-charge for mutually-exclusive audiences. Full rationale + Phase 2 trigger-gated additions in ops/ads/briefs/2026-05-20-targeting-stacks.md.

Phase 2 — trigger-gated additions

Stack	Trigger	Status check
LK_Prospect_2026Q2_lal_atc — LAL from ATC seed	LK_ATC_30d size ≥ 1k	Weekly via meta-ads wrapper
LK_Prospect_2026Q2_lal_purchase — LAL from PURCHASE seed	≥ 100 Pixel Purchase events / 180d (currently 4 lifetime)	Monthly
LK_Prospect_2026Q2_bodas_events — weddings/events reserve	Wedding season Mar–Sep OR Phase 1 plateaus	Activate manually
LK_Prospect_2026Q2_verano — summer/festivales	2026-06-15 + at least 2 summer-themed ads shipped	Calendar-gated
LK_Prospect_2026Q2_value_lal — value-based LAL	Customer file (email + value) ≥ 1k rows	Defer indefinitely

Specialist sign-off needed before launch. Billy validates: (1) Spanish interest names map to Meta taxonomy, (2) size floors hold after exclusions, (3) budget split €4/€4 vs consolidate to one €8/day ad set, (4) ship Ad set 3 at T+0 or defer to T+7.

Prospecting slate — 6 ads from Carla batch 1 (refs 3–7)

Each of the 6 ads (A1–A6) below runs inside every active ad set above. Per-ad performance reads cleaner because the creative variable is held constant across audiences.

Campaign B — LK_Retarget_2026Q2 · €4/day

Bonus prospecting pipeline (T+7 rotation candidates)

Carla batch−2 leftovers after fixes. Hold for the 2026-05-26 T+7 review — rotate in as variant freshness against the slowest-performing Campaign A ads. All need the same 3 systemic fixes from ops/ads/briefs/2026-05-20-carla-pushback.md.

Carla cycle (sent 2026-05-20, awaiting return)

Setup walkthrough — Ads Manager upload

1. Phase 0 · Audience hygiene (~10 min)
   • Rename existing 4 audiences with LK_ prefix: PAST_PURCHASE (30 DAYS) → LK_Past_Purchase_30d; ADD TO CART (30D) → LK_ATC_30d; ENGAGED_IG(90D) → LK_IG_Engagers_90d; ENGAGED_FB(90D) mark as deprecated (below floor).
   • Create LK_Past_Purchase_180d (Custom audience → Website → event = Purchase, retention = 180d).
   • Create saved audience LK_Retarget_Union = LK_IG_Engagers_90d ∪ LK_ATC_30d ∪ LK_VC_NotIC_14d ∪ LK_Site_Visitors_30d, excluding both past-purchase audiences.
   • Verify via meta-ads wrapper: 9 audiences with LK_* prefix, LK_Past_Purchase_180d Ready-for-use.
2. Phase 1 · Campaign A launch (Sales / Purchase / €8/day)
   • Create campaign LK_Prospect_2026Q2 — objective Sales, ad-set budget (no CBO).
   • Single ad set: name LK_Prospect_2026Q2_broad_es18-40. Settings per the Campaign A card above. Set the 4 exclusion audiences.
   • Upload 6 ads (A1–A6). For each ad: Advantage+ Creative enhancements OFF on every toggle (Music, Image animation, 3D animation, Visual touch-ups, Brightness/contrast, Image expansion, Image templates, Add overlays, Show product tags). Standard Enhancements OFF. Multiple text optimisations OFF. Sitelinks OFF. Language: Spanish (Spain) specifically.
   • Per ad upload Headline (≤27 chars), Primary text (50–150 chars), CTA = Comprar ahora, Destination = lucernaskin.com/products/cubrepezones with UTM utm_source=meta&utm_medium=cpc&utm_campaign=LK_Prospect_2026Q2&utm_content=A1 (etc).
3. Phase 2 · Campaign B launch (Sales / Purchase / €4/day) — blocked on Carla return
   • Create campaign LK_Retarget_2026Q2. Same Sales objective, separate budget.
   • Single ad set: LK_Retarget_2026Q2_union. Audience = LK_Retarget_Union saved. Exclude both past-purchase audiences. Placements + per-ad enhancements identical to Campaign A.
   • Upload 7 placement units once Carla returns: Angle 1 V-A + V-B (locked, ready now), Angle 2 V-A + V-B, Angle 3 V-A + V-B, Angle 4 carousel (3 cards 1:1, single ad).
4. Phase 3 · Reporting preset (~5 min, anytime)
   • Build column preset LK_Funnel_CR1-CR3: CR1 = Link Clicks + Link CTR · CR2 = Adds to Cart + Checkouts Initiated · CR3 = Purchases + Cost per Purchase + Purchase ROAS · plus Spend, Impressions, Frequency, Quality Ranking, Engagement Rate Ranking, Conversion Rate Ranking. Save and set as default for both campaigns.
5. Phase 4 · T+7 review (2026-05-26)
   • Kill bottom-2 ads per campaign by CPA outlier OR CTR < 0.5% after ≥1k impressions. Brief 2 replacement variants per campaign (single-dimension changes only). Bonus pipeline (refs 9 / 10 / 12) ready to slot in as Campaign A replacements.

Targeting — locked 2026-05-21 Upload-ready

Supersedes the v1 stacks doc and the menu framing. Two ad sets ship at T+0 under Campaign A — same 6 ads (A1–A6) run in each. Stacks 3–6 are pre-specified with trigger conditions, not "decide later". This section is the canonical upload spec; the markdown file (ops/ads/briefs/2026-05-21-launch-targeting.md) is a mirror.

Campaign A — LK_Prospect_2026Q2 envelope (applies to all ad sets)

Objective	Sales · optimise for Purchase
Budget control	Ad-set budget (no CBO) · Lowest cost bidding
Attribution	7-day click / 1-day view
Geo	Spain (peninsular + Baleares + Canarias)
Language	Spanish (Spain) — not the generic "Spanish" toggle (spills into LATAM)
Gender	Women
Placements	Manual: Feed (FB + IG) + Reels + Stories. Drop: Audience Network, Marketplace, Right column.
Detailed Targeting Expansion	OFF (every ad set, invariant)
Advantage+ Audience	OFF (every ad set, forever)
Total Campaign A budget	€8/day at T+0 (€4/€4). Scales per trigger schedule below.

Ad set 1 · LK_Prospect_2026Q2_lingerie_adj · €4/day · Age 22–40

Wedge: already pays €20–50 for invisible/shaping garments — AOV pre-qualified.

#	Interest (upload-name)	Meta ID	Audience size
1	Spanx	6003090472301	3.6M–4.2M
2	SKIMS	hand-verify in UI	—
3	Wolford	6003391746154	323k–380k
4	Calvin Klein Underwear	hand-verify in UI (Graph returns Calvin Klein parent — too broad)	—
5	Intimissimi	6003247379299	11M–13M
6	women'secret	6003102473578	2.2M–2.5M
7	Oysho	6009374462295	7.8M–9.2M
8	TEZENIS	6003170033702	3.6M–4.2M
9	La Perla NEW	6002991665859	522k–613k
10	COSABELLA NEW	6003148971549	778k–915k

Behavior overlay: Engaged Shoppers 6071631541183 · Diff vs v1: dropped Pretty You London + Étam (smaller / less premium); added La Perla + Cosabella (verified premium-AOV signal). Expected CPM: €6–12. Estimated reach (Meta will confirm at upload): 2–4M.

Ad set 2 · LK_Prospect_2026Q2_fashion_premium · €4/day · Age 18–40

Wedge: style-led shoppers at the premium-but-not-luxury fast-fashion band — where "vestido sin espalda" lives.

#	Interest (upload-name)	Meta ID	Audience size
1	Massimo Dutti	6003779753852	9.2M–10.8M
2	Mango (clothing)	6003132162739	46M–54M
3	Zara Woman	hand-verify in UI (Graph returns generic Zara parent — too broad)	—
4	Sandro	hand-verify in UI (Graph returns near-zero variant — brand exists, normalization issue)	—
5	Ganni	6003038366329	677k–797k
6	Reformation	6003013562293	1.4M–1.7M
7	COS	6012762926489	2.9M–3.4M
8	BIMBA Y LOLA NEW	6012367282363	1.6M–1.8M
9	Sézane NEW	6008848678164	1.3M–1.5M
10	Maje NEW	6003146996868	1.4M–1.7M

Behavior overlay: Online shopping 6003346592981 · Diff vs v1: dropped Free People (US-skewed) + "Vestidos de cóctel" + "Moda femenina" (failed verification); added Bimba y Lola + Sézane + Maje. Expected CPM: €5–10. Estimated reach: 4–6M.

Trigger schedule

When	Condition	Action
T+7 · 2026-05-28	Per ad set: CTR < 0.5% at ≥1k impressions	Kill underperformer, replace with Stack 4 (Bodas) at €4/day. Pre-specified below.
T+14 · 2026-06-04	Campaign A has ≥10 Purchases in trailing 7d	Add Stack 5 (Verano) at €4/day — festival season is live. Total Campaign A €12/day.
If T+0 both fail	Both ad sets < CR1 thresholds	Pause Campaign A — not an audience problem. Rework creative.
Held indefinitely	Campaign A plateaus AND need reach AND can't activate Stack 4/5 first	Stack 3 (Beauty) · AOV-analogue argument weak for body-product specifically.
Held indefinitely	Same as above + Stack 3 also exhausted	Stack 6 (Body confidence / media) · high-engagement low-purchase risk.

Stack 4 · LK_Prospect_2026Q2_bodas_events · trigger-ready (T+7)

Wedge: highest-intent narrow stack. Wedding/engagement/formal-event life-signal correlates directly with backless/strapless/halter dress purchase. Geo Spain · Women 25–38 · es_ES · same envelope settings as Ad sets 1/2.

#	Interest	Meta ID	Audience size
1	Wedding planning	6003069898229	5.1M–5.9M
2	Gala	6003616498687	4.8M–5.6M
3	Cocktail party	6003320213294	2.4M–2.9M
4	Vera Wang	6003381526743	1.7M–2M
5	Pronovias	6003282869002	1.2M–1.4M
6	Bodas.net	6003359354221	438k–515k
7	Rosa Clará	6003202087086	353k–415k
8	Monique Lhuillier	6003257275301	248k–292k
9	Vestido de novia	hand-verify in UI	—
10	Despedida de soltera	hand-verify in UI	—

Behavior: Engaged Shoppers 6071631541183

Life events overlay: Newly engaged (1 year) 6003050210972 · Newlywed (3 months) 6013133420583 · Anniversary within 30 days 6017476616183

Stack 5 · LK_Prospect_2026Q2_verano · trigger-ready (T+14)

Wedge: seasonal — Spanish festivales, beach holidays, resort wear. Activated mid-June through August. Geo Spain · Women 18–32 · es_ES.

#	Interest	Meta ID	Audience size
1	Mykonos	6003222090276	6.9M–8.1M
2	Calzedonia	6003326031941	5.6M–6.6M
3	Crucero	6015068466099	4.5M–5.3M
4	Formentera	6003697967183	1.6M–1.9M
5	Primavera Sound	6003372662140	891k–1M
6	Ushuaïa Ibiza Beach Hotel	406865796146660	759k–893k
7	Bilbao BBK Live	6005430973159	294k–345k
8	Festival Internacional de Benicàssim (FIB)	6003288316351	140k–165k
9	Arenal Sound	6003335987448	111k–130k

Behavior: Beach Travellers 6014648415813 (narrower than Frequent Travelers — better fit for AOV band)

Life events (optional, skip unless under-pacing): Away from family 6003053857372 · Away from hometown 6003053860372

Phase 2 · LAL audiences (trigger-gated separately)

LAL ad set	Seed	Status / trigger
LK_Prospect_2026Q2_lal_ig_engagers	1–3% LAL from LK_IG_Engagers_90d (9.5–11k, exceeds 1k min)	Ready to build now — deferred only because T+0 budget is fully allocated. Activate at T+14 alongside Stack 5 if scale gate clears.
LK_Prospect_2026Q2_lal_atc	1–3% LAL from LK_ATC_30d	Trigger: ATC audience size ≥1k members. Status-check weekly via meta-ads wrapper.
LK_Prospect_2026Q2_lal_purchase	1% LAL from Purchase seed	Trigger: ≥100 Pixel Purchase events / 180 days (currently 4 lifetime — far from threshold). Monthly check.
LK_Prospect_2026Q2_value_lal	Value-based LAL from customer file	Trigger: same as LAL Purchase + customer-file upload built.

Standard exclusions (every ad set in Campaign A)

• LK_Past_Purchase_30d + LK_Past_Purchase_180d (once built) — never advertise to known buyers.
• LK_ATC_30d — push to Campaign B retargeting instead.
• LK_IG_Engagers_90d — also pushed to Campaign B (Union audience).
• Detailed Targeting Expansion OFF on every ad set (invariant).
• Advantage+ Audience OFF on every ad set (forever — per [[feedback-advantage-plus-off]] memory).
• All LAL audiences (when activated) excluded from each other (1% LAL is strict subset of 3% — Meta does not auto-deduplicate).

Per-ad enhancements (every ad in Campaign A)

All toggles OFF (matches 2026-05-19-campaign-structure.md §4):

• Advantage+ Creative enhancements — OFF on every toggle: Music, Image animation, 3D animation, Visual touch-ups, Brightness/contrast, Image expansion, Image templates, Add overlays, Show product tags.
• Standard Enhancements — OFF.
• Multiple text optimisations — OFF.
• Sitelinks — OFF.
• Language — Spanish (Spain) specifically, not generic "Spanish".

Why this lineup (ranked)

Rank	Stack	Launch slot	Confidence	Reason
1	Lingerie adjacency	T+0	High	Cleanest AOV-band overlap. €20–50 lingerie buyer → €28.99 product is a direct AOV step, not an analogue. Best verified IDs (Spanx, Cosabella, La Perla, Intimissimi).
2	Fashion premium	T+0	High	"Vestido sin espalda" use-case + editorial register match. Spanish premium fast-fashion buyer (Bimba y Lola, Massimo Dutti, Mango) is the volume play.
3	Bodas / events	T+7 trigger	High	Highest intent but narrowest reach + most CPM-volatile. Held back so it doesn't compete with Stacks 1/2 for budget at T+0. Verified bridal + life-event coverage is excellent.
4	Verano / festivales	T+14 trigger	Medium-high	Seasonal — festival season ramps June 1. Holding for scale gate so spend isn't wasted while T+0 stacks are still in learning.
5	Beauty / skincare	Indefinitely	Medium-low	AOV-analogue argument weak for a body product specifically. Only 5 verified high-confidence brands.
6	Body confidence / lifestyle media	Indefinitely	Low	Softest wedge. Premium-media readers are high-engagement, not necessarily high-purchase on intimate-apparel-adjacent.

Why €4/€4 not €8 on one or €2.67×3: €8 on one stack kills the test (no comparison data). €2.67 across 3 ad sets is below Lowest-cost-bid's learning floor (consensus €4–5/day) — all three stay in learning indefinitely. €4/€4 across 2 stacks is the cleanest A/B with enough budget per ad set to exit learning.

On hand-verify rows: a handful of universally-known brands (SKIMS, Calvin Klein Underwear, Zara Woman, Sandro, plus the ES wedding category terms Vestido de novia / Despedida de soltera / Vestido de cóctel) don't surface cleanly via Meta's public Graph adinterest endpoint — either empty or matched to a too-broad parent. They exist in the Ad Set UI autocomplete. Type the name directly in the in-app search at upload — it will find them. Normal Meta behavior, not a data issue.

Canonical spec: ops/ads/briefs/2026-05-19-campaign-structure.md · Launch checklist: ops/ads/briefs/2026-05-19-launch-checklist.md · Carla pushback brief: ops/ads/briefs/2026-05-20-carla-pushback.md · Carla refs: ops/ads/creative/carla-refs-2026-05-19/ · Rendered assets: ops/ads/creative/drafts-2026-05-19/ · Targeting (locked): ops/ads/briefs/2026-05-21-launch-targeting.md · Verification receipts: ops/ads/briefs/2026-05-21-targeting-research.md · Plan file: ~/.claude/plans/session-handoff-jiggly-abelson.md.