Email and WhatsApp retention for an Indian ecommerce store

Most Indian stores we look at have a well-run acquisition setup and close to nothing behind it. Ads bring a customer in, the order ships, and the relationship ends there until that same person sees another ad and gets paid for twice. The list sitting in the database is the cheapest audience the business owns, and nobody is sending it anything except invoices.

Retention is operations work more than it is creative work. Here is how the pieces fit.

Why the economics of a repeat order are different

Every new customer has to be found and paid for, through an auction where the price is set by everyone else bidding for the same attention. A repeat order goes to someone who already handed over an address, already trusted you with a payment, and already knows whether the product was any good. You are not buying that attention at auction. You are sending a message to a list you own.

The second advantage matters more than the cost one. You know what this person bought, when, and whether it came back. That history lets a message be specific in a way no ad targeting can match, because it comes from your own order table and not a platform's guess about who someone resembles.

Email and WhatsApp are not interchangeable channels

They get discussed as two options for the same job. They work nothing alike underneath.

• Email is open-ended. You compose what you like and send it to whoever is on the list, and the real constraint is sender reputation: complaints, bounces and spam placement accumulate against your domain.
• WhatsApp business messaging is template-based. A message your business initiates has to follow a template submitted for approval in advance, so you cannot improvise copy the morning of a sale the way you can with an email.
• WhatsApp requires opt-in before a business can message someone at all, and it is an identified channel: the customer sees a business profile and can block it in one tap, with no arguing.
• Once a customer replies, the conversation opens up for a window, and most stores never use it. Support, order questions and returns live naturally there.
• Delivery on WhatsApp is near certain and gets read fast. Treat that as a reason to send less through it, not more.

Plan around the template approval. It belongs on the campaign calendar and not on the launch day.

Keep transactional and promotional messaging apart

Transactional messages go to a person because of something they did: an order confirmation, a shipping update, an OTP, a refund notice. Promotional messages go to a person because they agreed to hear from you about offers. The two have different audiences, different consent and, on email, they deserve different sending identities.

Mixing them is the most common self-inflicted wound here. Slip an offer banner into an order confirmation and you take a message people always open and start training them to treat it as marketing. Some will mark it as spam, and a complaint against the address that also sends order confirmations puts your receipts at risk. On WhatsApp the same mistake costs more, because the response to an unwanted promotion is a block, and a blocked number cannot deliver that customer's shipping updates either.

Order status is the highest-trust message you will ever send

People open shipping updates, and on delivery day they open them more than once. For a store running COD this message also confirms an address and a delivery window, which is real operational work: a customer who reads and confirms is less likely to refuse the parcel at the door.

Protect that. Accuracy first, and keep commercial content restrained enough that nobody rereads the message wondering whether it is an ad. A care instruction, a reorder link or a support contact sits fine inside a delivery message. A discount code does not, and the cost of testing that is a channel people stop trusting.

Segments you can act on, and segments that only look like segments

Recency and purchase history carry most of the weight. Everything else is usually decoration.

• Bought a consumable and has passed the point where a normal person would have run out. The most useful segment a store can build, and it needs only an order date and a product.
• Bought once, some time ago, never came back. A different message from the one you send a regular.
• Bought repeatedly in one category and never touched an adjacent one you stock.
• Started a checkout with a saved address and stopped. Time-sensitive, and the window is short.
• Returned or refused a delivery. Worth its own handling, sometimes worth no message at all.

The test for any proposed segment is simple. Name the message that segment gets, then ask whether that message would be wrong to send to everybody else. If the answer is no, you built a filter and called it a segment. City, gender and "engaged subscribers" fail this test constantly, because the store has no differentiated thing to say to any of them.

Consent and unsubscribe are operational requirements

An unsubscribe has to work on the first click, everywhere, immediately. That means one suppression list every sending tool reads, including the one your agency uses and the one somebody set up for a single campaign long ago. Someone who opted out and hears from you anyway will report you, and those reports cost you delivery for the whole list.

Record where each consent came from and when. A checkout checkbox, a popup, a WhatsApp opt-in during a support chat: log the source alongside the contact. That lets you answer a complaint with a fact instead of a guess, and lets you drop a list of dubious provenance without wondering what went with it. Numbers scraped or bought belong nowhere near this system.

Measuring retention without borrowing a benchmark

Industry averages for repeat rate are close to meaningless, because they blend categories with wildly different replenishment cycles, price points and purchase frequencies. A store selling something people buy monthly cannot share a target with one selling something people replace twice a decade. Measure against your own history.

1. Group customers by the month of their first order and follow each group forward. A cohort makes changes visible that a blended number hides.
2. Track the share of each cohort placing a second order, and how long that took. Second order is the threshold that matters, because the jump from one to two is where most customers are lost.
3. Track revenue per customer over a fixed window, ninety days or a year, held constant so the comparison stays honest.
4. Compare one cohort against the cohort before it. That comparison is yours, verifiable, and immune to whatever a vendor's benchmark report claims.
5. Watch unsubscribes and blocks as a rate alongside the wins, because a campaign that produced orders while burning a chunk of the list has not really succeeded.

A practical note. Length discipline matters in both channels: our word counter keeps subject lines and template bodies inside sensible limits, and the text extractor pulls every address or URL out of a pasted block when you are auditing an old list.

The store that sends four careful messages a year to a list that trusts it will outlast the one sending four a week to a list that has learned to ignore the sender name.

Should an Indian store use email or WhatsApp for retention?

Both, for different jobs. WhatsApp suits short, time-sensitive, high-trust messages such as order and delivery updates, and it demands opt-in plus approved templates. Email suits longer content, catalogue and offer messaging, and costs nothing per extra send. Keep transactional and promotional streams separate in both, and honour every opt-out immediately.