- Platform "cost per purchase" undercounts CAC: it ignores returning-customer orders it claims, RTO/COD failures, and spend on channels it can't see.
- True new-customer CAC = total acquisition spend ÷ net new customers who kept their order. In India, RTO alone can move CAC 20–30%.
- Lower CAC by fixing the funnel and the product mix before touching bids — the biggest lever is usually the landing page, not the auction.
Ask three people at the same brand for CAC and you'll get three numbers: the media buyer quotes Meta's cost-per-purchase, the growth lead divides monthly spend by monthly orders, and finance divides spend by new customers net of returns. Only the third number can run a business — and in India it diverges from the first more than anywhere else, because of one three-letter word: RTO.
The honest formula
- Total acquisition spend = all paid channels + agency/tool fees attributable to acquisition. Not just Meta.
- New customers = first-ever paid order. Strip returning buyers — platforms happily claim them.
- Net = subtract RTO (return-to-origin) and refunded first orders. A customer who never accepted delivery isn't a customer.
Worked example: the ₹612 CPA that was really ₹1,050
A brand spends ₹6,00,000 in a month. Ads Manager reports 980 purchases → a proud ₹612 cost-per-purchase. Now the corrections:
| Adjustment | Effect | Running count |
|---|---|---|
| Platform purchases claimed | — | 980 orders |
| Remove returning customers (say 22%) | −216 | 764 new-customer orders |
| Remove RTO on COD orders (say 25% of the 60% that chose COD) | −115 | 649 kept orders |
| Remove first-order refunds (3%) | −19 | 630 net new customers |
₹6,00,000 ÷ 630 = ₹952. Add the agency retainer and tools apportioned to acquisition (₹60k) and you're at ≈ ₹1,048 — 71% above the dashboard number. Every decision made at "₹612" (scaling budgets, hiring, pricing) was made against a fiction.
The four India-specific traps
1 · RTO is a CAC problem, not a logistics problem
COD share in tier-2/3 India is still huge, and COD RTO commonly runs 20–30%. You paid full acquisition cost for those orders and got inventory back with shipping burnt both ways. If your CAC math doesn't subtract RTO, your "profitable" campaigns may be your best RTO generators — cheap clicks often correlate with low purchase intent.
2 · Returning buyers inflate every platform number
Retargeting audiences are full of people who already know you. Platforms claim those orders at full credit. Your blended cost-per-purchase looks great while your new-customer CAC quietly climbs — the exact failure mode that makes brands feel efficient while growth stalls.
3 · GST and gateway fees hide in the gap
Ad spend is quoted ex-GST; you pay 18% on top (input credit helps, but cash flow feels it). Payment gateways take ~2%. Neither appears in any ROAS anywhere.
4 · Marketplace spend blurs the blend
If you also spend on Amazon/Flipkart ads, that spend acquires their customer, not yours — no email, no retention lever. Blending marketplace and DTC spend into one CAC hides that difference in kind.
How to actually lower CAC (in order of leverage)
- Fix the landing page before the auction. In audits we run, click→page rates are healthy (60–75%) while page→purchase sits at 1–2%. A one-point lift there beats any bid strategy ever invented.
- Prepay incentives beat RTO insurance. A 5% prepaid discount that shifts COD share from 60% to 40% often saves more than it costs, purely in RTO.
- Kill the money pits weekly. Most accounts have one campaign burning 20% of spend at 3–4× blended CPA. It survives because nobody joins spend to net orders weekly.
- Concentrate on the proven product. Spreading spend across the catalogue "for fairness" funds losers. Scale the SKU that converts; fix the others' pages first.
- Feed clean signal back. Server-side conversion data (net of RTO if possible) improves the platform's own bidding — the only way "the algorithm" gets smarter about who actually keeps orders.