What Amazon’s Checkout Teaches a Small Online Store (and What It Doesn’t)

The Baymard Institute, which has studied e-commerce checkouts for over a decade, puts the average documented online cart abandonment rate at 70.22%, calculated from 50 studies. Put simply, for every ten shoppers who add something to a cart, about seven leave without buying.

A typical owner I speak to runs a small online store from Thane or Navi Mumbai, selling clothing, home décor or specialty food. Their dashboard shows a few hundred carts a month and a fraction of that in orders. They look at Amazon and assume the gap is about budget and brand. Some of it is. But a large part of Amazon’s conversion comes from checkout decisions that any store can copy. This article answers one question: which parts of Amazon’s checkout should a small Indian store borrow, and which parts would be a mistake to copy?

1. Why checkout is where small stores lose the most sales

Some abandonment is normal. Plenty of people add items to compare prices or save them for later. Baymard separates these out: in its survey of US shoppers who had abandoned a cart, 42% said they were just browsing. The useful list is what the rest said went wrong during checkout itself.

Source: Baymard Institute, Cart Abandonment Rate statistics. Survey of US online shoppers; respondents could give more than one reason. US data, not India-specific.

Read that list next to Amazon’s product and checkout pages and you will notice something. Almost every item on it is something Amazon has visibly worked on: delivery dates shown before you add to cart, saved addresses and payments, a return policy line near the price, reviews on every product. That is the real lesson: Amazon’s checkout reads like someone went down a list like this one and removed each reason to leave, one at a time.

Baymard also estimates that large e-commerce sites could see a 35.26% increase in conversion rate by fixing checkout usability issues alone. That is an estimate for large sites, not a promise for yours, but it shows how much is lost after the shopper has already decided they want the product.

2. Lesson one: show the full price before the last step

Extra costs are the biggest single reason in Baymard’s data, at 40%, and another 12% left because they could not see the total cost up front. Those are two sides of the same problem: a shopper decides to buy at one price and then meets a higher one at the end.

On Amazon India, the product page usually tells you before you add to cart whether delivery is free or what it costs, and roughly when it will arrive. You are rarely surprised at the payment step. A small store can do the same without Amazon’s logistics.

• State your delivery charge on the product page, not just at checkout. “Free delivery above ₹999, otherwise ₹60” in one line is enough.
• If you charge for cash on delivery, say so before checkout, not as a surprise line at the end.
• Show the full order total, including GST and delivery, in the cart, before the address step.
• If delivery cost depends on pincode, add a pincode checker on the product page so people find out before they commit.

My opinion: a slightly higher product price with free delivery usually feels better to shoppers than a lower price with a delivery fee added at the end, even when the total is the same. Test it on your best sellers rather than taking my word for it.

3. Lesson two: remove fields, not just steps

Owners often ask me whether a one-page checkout is better than a multi-step one. Baymard’s research suggests that is the wrong question. What matters more is how many fields a shopper has to fill in.

Source: Baymard Institute, Checkout Optimization: Minimize Form Fields, based on its e-commerce UX benchmark. Global benchmark of large sites, not India-specific.

The fields I most often remove from Indian store checkouts are a separate billing address (default to “same as delivery”), “address line 3”, a mandatory email when you already have a phone number for order updates, a “company name” box on a consumer store, and a separate confirm-email field.

Make Indian addresses easier

Indian addresses are long and irregular: flat number, building, society, sector or road, landmark. Two things help a lot. Ask for the pincode first and fill in city and state from it automatically. And give a single, generous “landmark” field rather than forcing people to split their address into boxes that do not match how they write it.

4. Lesson three: let people buy as guests

In Baymard’s survey, 18% of shoppers abandoned because the site wanted them to create an account. Amazon can ask everyone to sign in because most of its customers already have accounts. Your store is not in that position. A first-time visitor has no reason to create a password for a shop they have never bought from.

Baymard’s research on account creation found that 42% of sites in its benchmark interrupt users before or during checkout to suggest creating an account. Its recommendation is to save that invitation for the order confirmation page. By then you already have the name, phone, address and order details, so creating an account needs only a password, and there is no risk the shopper thinks registration is required to buy.

• Make “Continue as guest” at least as visible as “Sign in”.
• Add a short note on the guest option: “You can create an account after your order.”
• On the confirmation page, offer account creation with three or four concrete benefits: order tracking, faster checkout next time, saved addresses.
• If you use phone OTP login, consider letting the OTP that confirms the order also create the account quietly, so there is no password at all.

See also: for a full list of fixes beyond checkout, read How to reduce cart abandonment on your online store.

5. Lesson four: put trust where the shopper hesitates

Nineteen percent of shoppers in Baymard’s survey left because they did not trust the site with card details, and 13% because the returns policy was not good enough. Amazon has a brand that does much of this work for it. A small store has to earn that trust on the page, at the moment the shopper hesitates.

Reviews, especially the first few

The strongest evidence I know on reviews comes from Northwestern University’s Spiegel Research Center, first published in 2017. Its analysis found the following.

Source: Spiegel Research Center, Northwestern University, “How Online Reviews Influence Sales”, first published 2017. US data, not India-specific.

Two points stand out for small stores. Most of the gain comes from the first five reviews, so getting every product from zero to five matters more than getting your best seller from 50 to 100. And a perfect 5.0 can look too good to be true. Do not hide honest four-star reviews.

Policies and payment cues

• Show your returns or exchange policy in one line near the price and again in the cart. “7-day easy exchange” is clearer than a link to a long policy page.
• Show the payment methods you accept, with the logos of UPI apps and cards, before the payment step.
• Put a real phone number or WhatsApp contact on the checkout page. For many Indian shoppers, knowing they can reach a person matters more than any badge.
• Use a well-known payment gateway, so the payment page itself looks familiar.

6. Lesson five: remember the customer, but don’t chase one-click

Amazon’s most famous checkout feature is 1-Click ordering. Amazon filed the US patent in September 1997, it was granted in 1999, and it expired in 2017. For two decades that patent kept the exact feature to Amazon and its licensees, which is part of why it became so associated with the brand.

I do not think a small store should try to copy one-click. It works for Amazon because customers already trust it, have saved cards and addresses, and buy from it repeatedly. A small store’s customers often buy once or twice a year. In India, payment also usually involves a step outside your site: approving in a UPI app, or confirming a card payment with an OTP. You cannot remove those, and you should not try.

What you can copy is the idea behind one-click: a returning customer should not have to type anything twice.

• Recognise returning customers by phone number and pre-fill their saved address.
• Remember their last payment method and put it first.
• Offer a “buy again” link in order confirmation and delivery messages for products people reorder, such as food, cosmetics or supplies.
• Keep the payment options most of your customers use at the top. For many Indian stores that means UPI first, then cards, and cash on delivery if you offer it.

7. Lesson six: make a delivery promise you can keep

One in five shoppers in Baymard’s survey abandoned because delivery was too slow. Amazon’s answer is speed backed by a huge logistics network, and you cannot match that. What you can match is clarity. Amazon shows an expected delivery date, not just a vague “ships soon”, and that certainty is a large part of its appeal.

• Show an estimated delivery window on the product page, ideally by pincode: “Delivered in 2 to 4 days to 400703.”
• If you deliver locally yourself, say so. “Same-day delivery in Navi Mumbai for orders before 2 pm” can beat Amazon for nearby customers.
• Be conservative. A promise of four days that arrives in three builds trust; a promise of two that arrives in four destroys it.
• Send order, dispatch and delivery updates on WhatsApp or SMS so customers do not have to ask.

8. Lesson seven: make it work on a cheap phone on a slow connection

Seventeen percent of shoppers in Baymard’s survey abandoned because the website had errors or crashed. In my experience, the errors that hurt Indian stores most happen on mobile: a pincode field that opens the wrong keyboard, a payment page that times out when switching to the UPI app and back, a “place order” button hidden under a chat widget, or a page that takes too long to load on a weak signal.

Amazon tests constantly across devices. You do not need its testing lab, but you do need to try your own checkout the way your customers use it: on a mid-range Android phone, on mobile data, with a real UPI payment.

• Use the right input types: numeric keypad for phone and pincode, email keyboard for email.
• Test the full UPI flow: leave the browser, approve in the UPI app, and come back. Make sure the order is confirmed even if the shopper returns slowly.
• Remove pop-ups and chat bubbles from the checkout page itself.
• Check page speed on mobile data, not office Wi-Fi.

See also: if your pages are slow, start with How to speed up your website.

Where this advice breaks down

I want to be clear about the limits of these lessons.

• The data is mostly American. Baymard’s abandonment survey covers US shoppers, and the Spiegel review study is US-based. Indian shoppers have different habits, especially around cash on delivery, UPI and price sensitivity. Use the figures to rank problems, not to predict your numbers.
• The review study is from 2017. Its broad pattern has been widely repeated, but fake reviews have grown since, and shoppers may be more sceptical of review counts now.
• Amazon’s advantages are partly structural. Prime members have already paid for delivery, and Amazon’s returns and logistics network cannot be copied. Some of its conversion rate comes from that, not from checkout design.
• Cash on delivery is a trade-off. It can raise conversion but also raises returns and refused deliveries. Whether to offer it, and whether to charge for it, depends on your margins and return rates.
• Small changes need enough orders to measure. If you get 30 orders a month, you cannot run a meaningful A/B test. Fix the obvious problems from the table above and judge results over a few months.
• Your platform may limit you. Some changes, like reordering checkout fields or remembering returning customers, are easy on one platform and need development on another.

Checkout checklist for a small online store

Seven out of ten carts are abandoned on average. You will never get that to zero, and Amazon has not either. But each reason on Baymard’s list that you remove brings back some shoppers who had already decided to buy, and those are the cheapest sales you will ever win.