D2C Skincare Brand Cuts Cost-Per-Purchase by 62% in 45 Days

The Challenge

SaffronGlow was a Pune-based D2C skincare brand selling dermatologist-formulated products — a Vitamin C serum, a niacinamide moisturiser, and a retinol night cream. The products were genuinely good; they had 4.6-star reviews and strong repeat purchase rates from customers who'd found them organically.

The problem was paid acquisition. Their Meta ad account was a disaster. They'd been running the same 3 static image ads for 8 months. Their campaign structure had 14 active ad sets all competing with each other. Their cost-per-purchase had ballooned to ₹1,200 — nearly 40% of their average order value of ₹3,100. They were essentially break-even on every new customer acquired.

Their founder told us: "We know our product works. We just can't figure out why our ads don't."

The Audit

Before touching anything, we spent 5 days doing a complete ad account audit. What we found was predictable but painful:

• Creative fatigue: The same 3 ads had been running for 8 months. Frequency was at 9.2 — meaning the average person in the target audience had seen the same ad 9 times. No wonder they'd stopped responding.
• No funnel structure: Every campaign was targeting cold audiences with a "buy now" CTA. There was zero retargeting. Warm audiences — people who'd visited the website, watched a video, or engaged with a post — were being completely ignored.
• Audience cannibalization: 14 active ad sets, many overlapping. They were competing with themselves, driving up CPMs and wasting budget.
• Wrong creative format: All static images in a Reel-first world. The algorithm was penalizing them with poor placements.

The Rebuild

We shut down the entire account and rebuilt it from scratch in 7 days. Three campaigns, each with a clear job:

• Campaign 1 — Awareness (cold audiences): Video ads showcasing the formulation story and real before/after results. Budget: 30% of total. Goal: video views and profile reach. We targeted skincare-interested women aged 22–40 in Tier 1 and Tier 2 cities.
• Campaign 2 — Consideration (warm audiences): Carousel ads retargeting website visitors, Instagram engagers, and video viewers. Each carousel highlighted a specific product with customer reviews as the hero copy. Budget: 30%.
• Campaign 3 — Conversion (hot audiences): Dynamic product ads retargeting cart abandoners and product page visitors with time-sensitive offers. Budget: 40%. This is where we concentrated the "buy now" messaging — on people who'd already shown purchase intent.

In parallel, we produced a full creative refresh: 2 UGC-style video ads (founder explaining the ingredients), 3 "skin transformation" carousel sets, and 4 static image ads with testimonial copy.

Week-by-Week Progress

Week 1–2: New campaigns launched. Campaigns in learning phase. CPP still high (₹980) but trending down. UGC videos outperforming statics by 3x on CTR.

Week 3: Retargeting campaigns kicked in. First wave of warm audience conversions. CPP dropped to ₹720. The niacinamide moisturiser was the clear bestseller — we shifted 60% of budget to its creatives.

Week 4–5: Hot retargeting campaign fully optimized. Cart abandonment recovery rate: 28%. CPP reached ₹540 by end of week 4. We introduced a "bundle offer" creative that became our top performer. ROAS crossed 3x for the first time.

Day 45: CPP: ₹456. ROAS: 3.8x. The account had been transformed.

The Results

• Cost-per-purchase dropped from ₹1,200 to ₹456 (62% reduction)
• ROAS improved from 1.8x to 3.8x
• Revenue from paid channels grew by 214%
• Cart abandonment recovery rate: 28%
• Monthly new customer acquisitions: 3x previous volume at lower cost
• Best-performing creative: UGC founder video — 4.2x ROAS

Key Takeaway

The most common mistake D2C brands make with Meta ads is treating the entire funnel as one stage. Cold audiences need to see, warm audiences need to consider, hot audiences need to convert. Running "buy now" ads to people who've never heard of you is like proposing marriage on a first date. The moment SaffronGlow started meeting customers where they actually were in their journey, the numbers fixed themselves.