My Nandu is a D2C e-commerce brand we ran paid growth for across 2024. The engagement landed at ROAS 12.2x and conversion rate around 3.5% against a Meta-primary funnel. Those numbers sound clean; the path to them was not. Here is the actual structure of the funnel, and the decisions behind it that are worth stealing.
The starting position
The brand had product-market fit locally but no organic distribution beyond word of mouth. Paid was running at ROAS 2.1x on a Meta account with eight active ad sets, most of which were serving near-duplicate creatives to overlapping audiences. The funnel was "run ads, land on product page, buy." There was no mid-funnel, no retargeting cohorts, no exclusion logic.
In other words, the account was spending money. It was not running a funnel.
What we changed, in order
Week 1: audit and exclusions. Before touching creative, we turned off seven ad sets that were cannibalising each other. Spend didn't drop; performance improved within seven days because the remaining ad sets got out of their own way.
Week 2 to 4: the three-tier creative structure. Top of funnel ran emotional hook videos, 9 to 15 seconds, testing one hook per creative against cold audiences. Middle of funnel ran product demonstrations, 20 to 30 seconds, targeting video view and engagement cohorts. Bottom of funnel ran testimonial plus offer, 30 to 45 seconds, targeting cart abandoners and product-page viewers.
The rule: every creative has a funnel stage, and spending budget on a creative that doesn't match its audience's stage is waste.
Week 5 to 8: the daily creative cadence. Two new top-of-funnel creatives every week. One new middle. One new bottom. Test duration: 72 hours minimum, 7 days maximum. Kill criteria: CTR below 1.2% or CPA more than 1.5x account average. This cadence produced the 12.2x ROAS within six weeks and held it.
The thing most marketers get wrong
Marketers obsess over creative. Creative matters. But the single highest-leverage lever on a Meta account is exclusion logic. Excluding recent purchasers from acquisition campaigns, excluding cart abandoners from cold reach campaigns, excluding retained customers from repeat-purchase pushes. These are not creative decisions. They are structural ones.
We audit exclusion logic before we look at a single ad.
The numbers that tell the story
- ROAS: 2.1x before, 12.2x after. 5.8x improvement.
- CVR: 1.4% before, 3.5% after. 2.5x improvement.
- CPA: directly halved once exclusions cleaned up audience overlap.
The CVR jump was not a landing page change. The landing page was unchanged. The traffic got better because the funnel decided who should see which ad at which stage.
Heuristics
- Audit the structure before you audit the creative. Exclusions, audience overlap, and ad set sprawl are where most accounts leak.
- Every creative has a funnel stage. Creative without a stage is creative you should not be running.
- Kill creatives on clean, pre-agreed thresholds. Emotional attachment to a creative that's underperforming is the single biggest budget drain in any paid account.
If you want the framing we used to scope the engagement before we touched a dollar of spend, that's here. The same one-page rule applies to paid as it does to product.
Written 2025-08-14 by Abhiraj Sakargaye.