What “where is my order?” really costs an Indian D2C brand

Updated 2026-09-24 · 5 min read · By the Convo360 team

Start with the number of messages, not the number of orders

Brands usually estimate WISMO by orders shipped. That undercounts it. One order produces a status question when it is late, another when the tracking page has not updated, a third when the customer forgot they ordered. A single delayed shipment can produce four messages across three days.

Pull a month of conversations and count how many contain an order word together with a status word. On most Indian D2C inboxes the honest figure lands between fifty and sixty-five per cent of all inbound messages.

Count the repeats separately while you are there. A customer who asks three times about one parcel is not three tickets of equal value, they are one failure that generated three conversations. Brands that track this usually find a small number of problem shipments producing a disproportionate share of the inbox, which points at the courier or the warehouse rather than at support.

Seasonality matters as much as the average. A Diwali week can carry three times the normal volume while your team is the same size and several people are on leave. Budget on the peak, not the mean.

Putting a rupee value on one reply

Take a support executive on ₹22,000 a month. That is roughly ₹127 an hour for 22 working days of eight hours, before any overhead. Now time the actual task: read the message, open the store admin, search, read the fulfilment state, copy the tracking link, write a human sentence, come back to the next chat.

Ninety seconds is a fair figure when the phone number matches on the first try. Two and a half minutes is fair when it does not.

Those numbers look small until you add the overhead nobody books: the supervisor, the tooling, the hiring, and the fact that nobody works at a hundred per cent occupancy. Loading it by two to three times gives the ₹12 to ₹25 range most founders recognise.

Two costs are usually left out of this sum and both are real. The first is context switching. An agent moving between the chat window and the store admin does not resume at full speed; studies of knowledge work put the recovery cost at tens of seconds each time, which is significant when the task itself is ninety.

The second is error. Copy the wrong tracking link into the wrong chat and you have created a second conversation, an apology and sometimes a privacy incident. It is rare per message and expensive when it happens.

The cost that does not appear in any spreadsheet

The real damage is queue position. A customer with a burst seam on a kurta, or a wrong size before a wedding, is sitting behind forty people asking where their parcel is. They wait four hours for a reply that should have taken ten minutes.

That customer is the one who writes the review. Measured properly, WISMO does not cost you salary, it costs you the response time on the messages that decide whether somebody buys again.

This is why first response time on complaints is a better health metric than average first response time across everything.

There is a revenue side too, and it is larger than the salary side. Repeat purchase rate is the number most D2C brands live on, and it is sensitive to how the last problem was handled rather than to how fast a parcel moved. A customer whose complaint waited four hours behind a queue of status questions remembers the wait, not the courier.

What changes when the bot answers

When the answer comes from your store automatically, three things move at once. The status question is answered in seconds at any hour. The agent's day empties of repetition. And the queue in front of a real problem shortens dramatically.

On Convo360 the bot reads the order from your connected Shopify store and replies with the real state and tracking link, and only ever to the customer who placed it. The setup is covered in letting the AI answer order questions.

A sensible target after a month: order questions answered without a human above eighty per cent, and first response time on everything else under fifteen minutes in working hours.

Be careful about what you promise internally. Automation does not reduce your Meta conversation bill, because the customer still writes and still gets a reply. What it reduces is people-time and waiting-time. If the business case was sold on message costs it will look like a failure even when it worked.

It also does not reduce the underlying failures. If fifteen per cent of parcels are late, fifteen per cent will still be late; you will simply hear about it faster and more calmly.

Do the sum for your own brand

Four numbers, fifteen minutes of work:

1. Inbound messages last month.
2. The share that mention an order and a status word.
3. Your loaded cost per agent hour.
4. Honest average handling time, measured by watching, not guessing.

Multiply them. Then ask a second question that matters more: how long did a customer with a genuine complaint wait last month, and what would that time be if two-thirds of the queue vanished?

If the answer changes how the week feels for your team, the automation is worth doing before the next festive peak, not after it. Brands that wait usually discover in October that the peak is exactly when the queue is longest and hiring is hardest.

Finally, run the sum again after ninety days with real figures rather than estimates. Handling time is almost always worse than the first guess, and self-service rates are almost always better. Both errors point the same way, which is why brands that measure honestly tend to expand the automation rather than roll it back.

Example conversation

Frequently asked questions

Related reading

Available across India — see WhatsApp & AI calling automation by state and city.