How We Cut Cost Per Purchase from AED 80 to AED 22: A Meta Ads Playbook
One of our e-commerce clients came to us paying AED 80 per purchase on Meta with an average order value of AED 180. After six weeks, cost per purchase was AED 22. Nothing about this was luck, and none of it required a bigger budget. Here is the exact playbook, in the order we executed it.
Week 1: Fix the Data Before Touching the Ads
The account was optimizing on bad data: the pixel double-counted some purchases and missed others entirely (roughly 30% of iOS conversions were invisible). We implemented the Conversions API alongside the pixel, deduplicated events, and verified every event in Events Manager against actual store orders.
Why this comes first: Meta's algorithm optimizes toward whatever data you feed it. Feed it 70% of reality and you'll pay 130% of the necessary cost.
Week 2: Consolidate the Account Structure
We inherited 14 campaigns and 41 ad sets, most spending under AED 50/day — every one of them stuck in learning phase, bidding against each other in the same auctions. We rebuilt to:
- 1 testing campaign — new creative concepts at controlled spend
- 1 scaling campaign — proven winners with consolidated budget (Advantage+ style, broad targeting)
- 1 retargeting campaign — site visitors, cart abandoners, and engaged social audiences
Within 10 days the scaling campaign exited learning phase for the first time in the account's history.
Weeks 2–6: Creative Testing at a Fixed Cadence
We launched 4 new creative concepts every week — not 4 variations of the same idea, but genuinely different angles:
- Problem-first UGC ("I couldn't find abayas that fit like this…")
- Offer-led statics (bundle + free delivery)
- Social-proof carousels (reviews, before/after)
- Founder story video
The winner wasn't what the client expected: a plain UGC video shot on a phone outperformed the polished studio content by 3.2× on ROAS. You don't know until you test — which is the whole reason our process exists.
Weeks 3–6: Match the Offer to the Funnel
Cold traffic saw a low-friction entry offer. Warm traffic saw the bestseller bundle. Cart abandoners got a 24-hour free-shipping nudge via retargeting. Same products, different doors — and every audience converted at meaningfully different costs. Blended CPP fell every single week for five consecutive weeks.
The Weekly Kill Rule
Every Monday, any ad whose 7-day CPP exceeded target by 50%+ was paused — no sentiment, no "let's give it one more week." Fatigued ads don't recover; they drag the account average down while you wait.
The Results, Summarized
| Metric | Before | After 6 Weeks |
|---|---|---|
| Cost per purchase | AED 80 | AED 22 |
| ROAS | 1.4× | 4.6× |
| Learning-phase ad sets | 41 | 0 |
Frequently Asked Questions
Can I copy this playbook myself?
The structure, yes. The hard parts are creative volume and disciplined weekly decisions — that's where most self-managed accounts stall. If you'd rather have it done for you, book a free audit and we'll map your version of this plan.
Does this work for services, not just e-commerce?
Yes — swap "purchase" for "qualified lead" and the same sequence applies. See our take on why ads don't convert for the lead-gen angle.
Ready to grow with a team that tests instead of guesses?
Get a free audit of your marketing. We will tell you exactly what to fix first — no commitment, no fine print.
Get FREE Consultation