Executive Summary

Before Meta announced Andromeda, this program was already running on the same principles: creative diversity, simplified structure, and a tight feedback loop between paid and retention. Every element — campaign design, audiences, creative cadence, Klaviyo flows, and measurement — was built and executed personally by Nicola Lewis. No agency. No outside support.

Context: Inventory Reality × Lockdown Legacy

Lockdown froze wholesale. Full inventory sat ready — archive pieces to clear, plus limited new runs. That allowed aggressive spend without stock-out risk and demanded disciplined creative rotation (swap, don’t pause).

A new audience also emerged: older women (40s–50s) drawn to the brand’s niche, rock-and-roll aesthetic. Early reach work produced first-party signal that seeded 1% lookalikes in 2023–24. CPMs were low, attention high, and ROAS consistently above typical apparel benchmarks.

Note: The wholesale business operated separately with its own marketing team. Results here reflect the direct-to-consumer campaigns and Klaviyo retention managed end-to-end by Nicola Lewis.

This campaign ran inside an ad account with around 12–15 other active campaigns but was the only one designed as a long form test — a nine-month lab to observe how Meta’s system learned when left to run without interruption.

The Human Signal Layer

Customer-service reviews showed buyers used exact product names (“Is the Alexandra back in stock?”). That language signalled memory and intent. We kept attention warm by replacing sold-out creatives rather than pausing, and mirrored customer phrasing in ad copy and flows.

“Most marketers never touch customer service. I did. Every complaint, every compliment, every ‘where’s my order?’ taught me who was really buying.”

Campaign Structure

• Objective: Conversions / Sales.
• Ad sets: 2 micro-niches (around 450k each), overlap manually removed.
• Creative volume: 35+ assets across video, carousel, UGC, lifestyle, and product-only; seasonal rotation.
• Budget split: 70% prospecting / 30% remarketing with capped bids.
• Placements: 27+ across Facebook, Instagram, Messenger (WhatsApp noted for future use; not active).
• Attribution hygiene: Offline Events mapped to sales for an extra validation layer.
• Integration: Direct Meta → Klaviyo handover so retention flows could convert and compound.

Results Snapshot

Retention & Klaviyo Performance

Meta discovery fed high-intent sign-ups into Klaviyo; 229 campaigns and always-on flows converted and compounded lifetime value.

• Meta → Klaviyo loop: newsletter sign-ups from ads converted through automated sequences.
• Attribution aligned: 7-day click windows kept measurement consistent across paid and email.
• Content echo, not copy: flows used new angles (fit, care, returns intel) instead of recycling ad creatives.

Data Sources & Signal Map

Where the truth came from — and how it shaped decisions.

Internal Validation

On 14 May 2024, Nicola emailed management summarising the live results nine months in. The message documented the same structure later formalised in Andromeda: persistent lifetime budget, micro-niche ad sets, 35 creatives on rotation, and a direct Meta → Klaviyo loop.

“I started one campaign to test my strategy back in Aug last year (still active)… two ad sets around 450k and 35 creatives constantly updated with new copy and media. Hence the importance of creative that works on socials.”

Peer Group Benchmarks (Fashion / Apparel)

“These results weren’t a team effort. They were mine alone — strategy, execution, testing, and analysis from start to finish.” — Nicola Lewis

Why It Matters

When Meta launched Andromeda in late 2024, it validated this method: creative as targeting, persistent budgets, and full-funnel feedback between acquisition and retention.

• Creative is targeting: variety trains the system faster than audience splits.
• Know before you scale: niche brands earn data through depth, not volume.
• ROAS is a vanity mirror: profit begins where attribution ends.
• Gut still wins: algorithms follow data; intuition spots change first.

Closing Note

Andromeda made the philosophy official. This campaign proved it first — built from scratch, by hand, with receipts.

Executive Summary

New customer acquisition has declined sharply while returning customers mask the drop in top-funnel performance. Audience overlap exceeds 50%, targeting remains all genders 18–65+, and attribution is inflated as Meta and Klaviyo both claim the same conversions. Once adjusted for new-customer only impact, Meta’s effective ROAS halves and true End CAC nearly doubles.

Meta charges on impressions. Broad targeting + high overlap = paying for non-prospects and duplicate reach.

Performance Snapshot (90 Days)

Detailed Analysis

Audience & Spend

• Major spend on 18–24 (£174) and 25–34 (£1,532) with weak purchase yield.
• 45–54 performs best (CPP £4.71) with strong volume.
• Reallocate toward 45–54 for ≈ +326 purchases without increasing total budget.

Attribution Inflation (Meta × Klaviyo)

• 40–50% of Meta-reported conversions overlap with Klaviyo. Shorten Meta attribution (1-day click/1-day view) and separate new vs returning in reporting.
• Estimated waste over the last 90 days: 30–40% of Meta spend.

The Takeaway

ROAS flatters. End CAC tells the truth. Without segmentation, fresh creative, or proper attribution, Meta becomes an expensive echo chamber for customers you already own.