Dealing with fake orders and time wasters

Three orders came in overnight. All cash on delivery, all ₹649, all to addresses you can't quite picture, and none of the three numbers is answering at eleven the next morning.

Almost none of that is fraud. The overwhelming majority of what small sellers call fake orders are real people who really did want the thing at 11:40pm and had completely stopped wanting it by Thursday, and the reason it hurts is not deception. It's that you paid freight out, freight back and a handling fee on an order that was never going to be collected.

So the job isn't detecting liars. It's confirming intent before you print a label, which takes about ninety seconds per order and is the single highest-return habit in cash-on-delivery selling.

Four kinds, and only one is malicious

Sort them, because the response is different for each.

The forgotten impulse. Ordered late, from a reel, with no real plan. This is the big one, and it's most of your losses by value. They're not lying. They've just moved on, and when the rider calls they'd rather not answer than say no.

The wrong details. Fat-fingered pincode, missing flat number, a phone number with nine digits. Genuinely wanted it, genuinely can't be reached.

The nobody's home. The person who ordered is at work. Whoever opens the door isn't handing ₹649 to a stranger for a parcel they know nothing about.

The deliberate fake. A competitor, a bored teenager, an ex. Real, rare, and disproportionately memorable, which is why sellers overestimate it. These usually give themselves away by being obviously wrong rather than subtly wrong: a huge order from a new number, an address that isn't a place, three orders in four minutes.

Count your own for one month before you build any system. Most sellers who count are surprised by how small the fourth category is, and how expensive the first one is.

What actually predicts a refusal

From sellers who track this, the signals worth acting on, roughly in order of usefulness:

Signal	What it means	What to do
No reply to the confirmation message within a few hours	The strongest single predictor there is	Hold. Follow up once. Cancel if silent
Ordered between midnight and 4am	Impulse, high refusal rate	Confirm before shipping, always
Address with no landmark, floor or house number	Rider can't find it, marks failed	Ask for a landmark before you ship
Number goes straight to a dead tone	Wrong number, not a busy one	Hold
Value well above your average basket	More cash to find at the door	Confirm, and consider a deposit
Multiple orders, same address, different names, same hour	The one that's actually suspicious	Call before shipping any of them
Pincode you've had refusals from before	Sometimes the area, usually the courier's local partner	Track it and decide

What is not a signal, whatever the internet says: a Gmail address, a female name, a rural pincode, a first-time buyer, or an order placed on a phone. Screening on those loses you real customers and it's the fastest way to build a business that only serves people who look like you.

The confirmation message, which is the whole system

Send it within an hour of the order landing. Not a template that reads like a template, and it must require a reply.

Hi Priya, thanks for the order. Confirming: 1x fast-charge cable, 1.5m, ₹649, cash on delivery to 214 Sindhi Colony, near the Ganpati temple, Jaipur 302001. Please keep ₹649 ready, the rider won't have change for ₹2000. He'll call from an unfamiliar number, so please pick up. Reply YES and I'll dispatch today.

Four jobs at once. It confirms they meant it, it confirms the address by reading it back, it warns them about the cash, and it pre-authorises the unknown number that would otherwise become a failed delivery attempt.

Then hold the line on what happens next:

• YES within a few hours. Ship it today.
• No reply by evening. One follow-up: "Still holding this for you, just reply YES when you're ready."
• No reply by the next evening. Cancel it and say so, warmly: "No worries, I've cancelled this one so it doesn't get dispatched by mistake. Order again any time and I'll sort it out."

A cancelled order costs you nothing. A dispatched order that comes back costs you two lots of freight, the handling fee, the packaging, and four to ten days with your stock in a van.

The seller detail that surprises people: the reply rate to that message is usually well over 80% among people who actually want the thing, and the ones who don't reply are almost exactly the ones who would have refused. You're not filtering by honesty. You're filtering by whether they still care enough to type three letters.

Make prepaid the cheaper option

The strongest lever available to an Indian seller isn't screening. It's pricing the risk into the two options and letting the buyer choose.

Take a ₹649 cable, ₹80 freight each way, ₹190 of product cost, and a COD handling fee on collection. If one COD order in four comes back, then across four orders you collect three times and pay return freight once, so every COD order carries roughly ₹20 to ₹40 of expected refusal cost before anyone refuses anything. That's the number you're allowed to give away to move people to prepaid.

So:

₹649 cash on delivery, or ₹599 if you pay now by UPI. Prepaid orders also go out the same day.

Fifty rupees off, plus faster dispatch, which costs you nothing and is the part buyers actually respond to. UPI to a plain UPI ID needs no gateway, no setup and no website, so this costs you nothing to offer beyond the discount itself.

Don't expect everyone to switch. COD is the default expectation in India and a shop with no COD option looks suspicious rather than efficient. But moving even a quarter of your orders to prepaid changes the arithmetic of the whole month, and the buyers who switch are disproportionately your repeat customers.

For anything above roughly three times your average basket, stop offering pure COD and take a deposit instead. The mechanics of that, including how to say it without losing the sale, are in taking a deposit before you start.

Keep a list of numbers

One line per refusal: phone number, date, order value, what happened.

When a number that's refused twice orders again, you don't refuse them. You change the terms:

Hi, I've got your order. I've had two parcels come back to this number before, so this one I can only send prepaid. ₹599 by UPI and it goes out today, here's the ID: [id]. If that doesn't work for you, no hard feelings.

Say it flatly, without accusation, and give them a real route to buy. About a third pay. The rest were never going to collect anyway. Being wrong about a genuine customer here costs you one sale; being wrong the other way costs you the parcel twice.

The wider question of drawing this line, and where it sits, is in when to say no to a customer.

The genuinely fake order

Occasionally you'll get one that's clearly hostile. A large order to a real address from a number that doesn't exist, or four orders in a minute, or an order placed with a competitor's shop name in the notes.

Do three things and then stop thinking about it.

Don't ship. No confirmation reply means no dispatch, and this is exactly the case that rule exists for.

Screenshot everything with timestamps, and keep it. If it's repeated and it's targeting you, you may want it later.

Don't post the number publicly. Ever. It's often someone else's number, entered by whoever placed the order, and the person you'd be exposing is usually another victim. Sellers who post "SCAMMER" with a phone number end up apologising to a stranger about once a year.

If someone is placing volumes of fake orders at you, the practical defence is turning off cash on delivery for a week. It's blunt and it works, because the whole point of the attack is that COD costs you money and costs them nothing.

Time-wasters in the DMs are a different problem

Not the same thing, and mostly not malicious either. Somebody asking "price?" fourteen times is not costing you freight, they're costing you attention, which is more valuable and harder to protect.

"Price?" under every post. Answer once, publicly, with the price and the link. Don't move to DM to give a price. A price given in a DM has to be given again to the next forty people.

The endless questioner. Three or four questions is a buyer. Eleven is a hobby. Close it politely:

Happy to answer anything else, but I think you've got everything you need to decide. The link's here whenever you're ready, and I'll hold one back for you until tomorrow evening if you want it.

A deadline is the kindest thing you can offer someone who can't decide. It ends the conversation either way, and it does it without rudeness.

The haggler. Have one answer and use it every time.

₹649 is the price, sorry. I do ₹1,199 for two if that helps, which is the only discount I run.

Never negotiate down in stages. A seller who moves from ₹649 to ₹600 to ₹575 in one conversation has taught the buyer that the real price is whatever they push for, and that buyer tells their friends.

The one who wants to chat. Genuinely enjoy it for two minutes, then leave the door open and go back to work. Some of these become your best customers, and none of them become customers if you're irritable.

If the volume of this is the actual problem, what to do when DMs get out of hand is about the systems rather than the wording.

Worked example: Rahul, phone accessories, Jaipur

Rahul sells cables, cases and chargers out of a small room in Jaipur. A fast-charge cable is ₹649, cases run ₹399 to ₹899, and he does around 60 orders a month, most of them from Instagram reels, almost all cash on delivery.

His third month was the bad one. Sixty orders, fifteen came back. That's 25%, which he later found out is roughly ordinary for COD in India rather than evidence that he was being targeted. At ₹80 each way plus ₹25 of packaging, those fifteen cost him about ₹2,775 in direct costs, and four of the returned boxes came back opened and unsellable.

Three changes, in this order.

The confirmation message, sent within an hour, ending in "reply YES and I'll dispatch today". He held anything unanswered and cancelled after 36 hours. Refusals went from 15 a month to 6 in the first full month.

Then the ₹50 prepaid discount with same-day dispatch. About a fifth of his orders moved to UPI, and those had a refusal rate of essentially zero.

Then the list of numbers. In seven months it has 23 entries, and nine of those numbers have tried to order again.

His month now runs around 70 orders with 4 or 5 refusals, and the thing he'd tell anyone starting is that none of it was a technology. It was asking people whether they meant it, and being willing to cancel the ones who didn't answer.

The bit that took him longest to accept: cancelling an unanswered order feels like throwing away a sale, and it isn't, because that sale was already gone. He just hadn't paid for it yet.

What Sailo does here, and what it doesn't

Very little, and you should know that before you build a process around it.

There's no phone verification, no OTP on the order form, no COD risk score, no address checking and no blocklist. Anyone who can open your shop page can place an order. There's a rate limit that stops a script firing off a thousand orders a minute, which is a defence against automation and no defence at all against one bored person placing three.

What you get is the order, with the items, the options, the address, the note and the total, sitting at unpaid until you mark it paid yourself, and a free-text delivery notes field on the cash on delivery rail that the buyer reads before ordering. Use that field for the two sentences that do the most work: where you deliver, and the fact that the rider carries no change.

Everything else, the confirming, the holding, the cancelling, the list of numbers, is manual and it's yours. That's the trade on a rail where the platform never touches the money and takes nothing from it. If you're running several hundred COD orders a month, India has checkout companies that do risk-score COD orders and will charge you for it, and at that volume they're probably worth the money.

Do this on the next order that lands

Write your confirmation message now, with your own item, price, address format and the line about change, and save it as a keyboard shortcut. Send it on every order for two weeks and count the refusals before and after.

Then add the delivery notes to your COD settings, and start the list of numbers today, even if it has one entry.

The full arithmetic of what a refused parcel costs, and how to get your money out of the courier once it does deliver, is in cash on delivery for small sellers. And most of this is downstream of something earlier: a buyer who trusts you enough to prepay never becomes a refused parcel, which is what building trust before money moves is about.