Every Indian D2C brand running cash on delivery eventually hits the same wall: a chunk of orders never make it to the customer's hands. The parcel goes out, the courier attempts delivery, nobody picks up or the customer says "I didn't order this" — and it comes straight back. That's RTO, and the single best defence against it is confirming the order before you ship.
The question isn't whether to confirm. It's how. There are three real options: an SMS or WhatsApp message, a human calling team, or an AI voice agent that phones the customer automatically. Each has a genuine place. This post lays out the honest trade-offs — cost, pickup rate, and scale — so you can pick the one that fits where your brand actually is.
The only metric that matters: cost per confirmed order
It's tempting to compare these three on price per message or price per call. That number lies. An SMS at 15 paise looks unbeatable until you notice that 8 out of 10 go unread. What you actually care about is cost per confirmed order — how much you spend to get a clear yes or no that you can act on before dispatch.
A channel that's cheap to send but rarely gets a response is expensive per confirmation. A channel that costs more per touch but reliably gets an answer can be far cheaper per outcome. Keep that lens on as we go through each one.
Option 1 — SMS and WhatsApp
The lowest-effort option, and the one most brands start with. You fire off a "Reply YES to confirm your order" message, sometimes with a link. It's dirt cheap, fully automated, and scales to any volume instantly.
The problem is engagement. Indian customers are drowning in transactional SMS and marketing WhatsApp. Most confirmation texts are ignored — not because the customer wants to cancel, but because a text simply doesn't demand a response the way a ringing phone does. You end up with a large pile of "no response" orders and no idea which are genuine buyers and which are RTO risks. WhatsApp does better than plain SMS when the customer already has your number saved, but the ceiling is still low.
As a first, cheap touch — or a follow-up nudge after a call. Great for prepaid order updates and shipping notifications. Weak as your primary COD confirmation layer, because silence isn't an answer.
Option 2 — A human calling team
The traditional answer: hire two or three tele-callers (or outsource to a BPO) who phone every COD customer and confirm the order by voice. It works. A real human is persuasive, handles odd questions, and can talk a hesitant customer into keeping the order. For high-value or consultative products, that human touch genuinely converts.
But the economics are brutal at scale. A single caller manages maybe 80–120 calls a day once you account for redials, breaks, and the fact that half the calls go unanswered on the first try. Between salary, telecom costs, training, attrition, and idle time, a human team lands somewhere around ₹12–25 per order confirmed — and that number gets worse, not better, as volume spikes during sales. You also can't call at 9pm or on a festival day without paying overtime, and quality drifts as callers get tired or leave.
Human teams don't scale linearly. Doubling your orders means roughly doubling headcount — and hiring, training and retaining tele-callers in India is a constant grind. A Diwali order spike is exactly when your team is most overwhelmed.
Option 3 — An AI voice agent
The newer option, and the reason this post exists. An AI COD confirmation call works like a human call — the phone actually rings — but it's placed automatically the moment a COD order comes in. The agent greets the customer by name, reads back the product and amount in Hindi, Hinglish or a regional language, and records a clear confirmed / cancelled / no-answer outcome that gets written straight back to your store as an order tag.
Because it's software, it does the two things humans can't: it scales to any volume instantly (a hundred simultaneous calls is the same as one) and it costs a fraction as much — typically ₹2–6 per call. It calls within minutes of the order, retries no-answers on a schedule, and never gets tired or quits. It won't out-negotiate a skilled human on a ₹15,000 order, but for routine COD confirmation it captures the vast majority of the value at a sliver of the cost.
The sweet spot is 300–3,000 COD orders a month — high enough that a human team is expensive and hard to staff, routine enough that a well-scripted voice call confirms the order without needing human judgement. That's most growing D2C brands in India.
Side by side
Here's how the three stack up on the dimensions that decide cost per confirmed order:
| SMS / WhatsApp | Human team | AI voice agent | |
|---|---|---|---|
| Cost per touch | ₹0.15–1 | ₹12–25 | ₹2–6 |
| Response / pickup rate | Low | High | Medium–High |
| Scales with volume | Instantly | Needs more headcount | Instantly |
| Speed to first contact | Instant | Hours (queue) | Minutes |
| Hindi & regional languages | Text only | Yes | Yes |
| Handles complex questions | No | Yes | Simple ones |
| Consistency & availability | Always | Varies, business hours | Always, within call window |
| Auto-tags the order in your store | Needs setup | Manual | Automatic |
Cost ranges are indicative for the Indian market in 2026 and vary with volume, language, and provider.
So which should you pick?
It genuinely depends on where your brand is — there's no single right answer:
- Under ~150 COD orders a month, or very high-value products: a human touch (yours or a small team) is fine and often best. The volume is manageable and the per-order value justifies the attention.
- 300–3,000 COD orders a month: an AI voice agent is almost always the right call. It gives you the reliability of a phone call at close to the cost of an SMS, and it doesn't buckle when orders spike.
- Very high volume with complex, consultative sales: a hybrid — AI confirms the routine majority, humans handle the high-value or flagged exceptions.
And SMS/WhatsApp isn't a competitor to any of these so much as a companion — the perfect cheap follow-up for the customers a call couldn't reach.
An AI confirmation call costs ₹2–6. Each RTO you prevent saves ₹180–350 in forward and reverse logistics. You only need to stop roughly 1 RTO per 60–100 calls to break even — and a real confirmation call clears that bar comfortably. See our complete Shopify RTO setup guide for how the whole flow fits together.
What good AI confirmation actually looks like
Not all "AI calling" is equal. A robotic IVR that barks "press 1 to confirm" gets cut in three seconds — customers in India have zero patience for it. What works is a natural, human-sounding voice that speaks the customer's language and keeps the call short. If you're wondering what to actually say, we broke down the exact wording in our COD confirmation call script for Hindi & Hinglish.
The other half is integration. The outcome of every call has to land back in your store automatically — a cod-confirmed or cod-cancelled tag — so your fulfilment team only packs orders that were actually confirmed. A call that doesn't update your store is just noise.
The takeaway
SMS is cheap but silent. A human team is reliable but doesn't scale and burns cash. An AI voice agent sits in the middle — the reliability of a real phone call at roughly the cost of a text, and it handles a festival-day order spike without you hiring a single extra person. For most growing Indian D2C brands, that's the option that actually moves RTO.
The best way to know is to hear it on your own store. Point CallFox at your Shopify store, let it call your next batch of COD orders, and watch the confirmed and cancelled tags show up in your admin — no calling team, no scripts to write, no code.
Try AI COD confirmation on your own store — free for 2 weeks.
CallFox calls your COD customers in Hindi and regional languages, confirms the order, and tags it in Shopify automatically. No setup fees, no calling team, no code.