Test offers at checkout

Kite works out whether your checkout loses money because shoppers leave or because the orders they place are small. Those two problems call for different offers, so Kite looks at both before proposing anything.

On a website Kite hosts, it reads the page side from its built-in analytics: how many people visit your product or sales page, and how many click your buy button. Stripe supplies the rest. Kite reads your checkout sessions to see how many shoppers who reached checkout went on to pay, how many let the checkout expire, and how much the completed orders were worth.

Those readings usually point one way. Suppose plenty of shoppers click "Buy" but a large share of checkouts expire unpaid. That suggests something at checkout, often a cost the shopper didn't expect such as shipping, is turning them away. Now suppose almost every checkout completes, but nearly every order contains a single item. Then the bigger opportunity is raising the order value, with a bundle or an add-on, rather than cutting the price.

Kite also uses what it knows about your products, your customers, and your competitors from its research on your business. If every competitor in your category offers free shipping above a certain order size, that's worth knowing before you test a discount instead.

Test the offer that answers the problem your numbers revealed, and test one offer at a time. If you add free shipping and a bundle in the same version, a better result can't tell you which one earned it.

These are the offers Kite tests most often:

• A percentage or fixed-amount discount, such as 10% off or $10 off, either applied automatically or as a code the shopper enters at checkout.
• Free shipping, either on every order or above an order size set just over what shoppers usually spend.
• A bundle, where two or three products that are often bought together sell as one item at a combined price.
• An add-on, a smaller product offered alongside the main one, such as refills, a case, or a gift wrap option.
• A spend threshold, such as $10 off orders over $75, which rewards shoppers for adding one more item.
• An offer after payment, where the confirmation page presents a related product with its own quick checkout.

Kite chooses among them using what it found in your data. A store whose checkouts expire unpaid usually starts with free shipping or a discount. A store whose shoppers buy one item and leave usually starts with a bundle, an add-on, or a spend threshold, because those offers are aimed at average order value rather than at getting the sale.

Every test Kite proposes starts as one sentence that states the evidence, the change, the expected effect, the shoppers it applies to, and how the result will be measured. For example: "Because 40% of checkouts expire unpaid and shipping is added at the last step, offering free shipping on orders over $50 will increase completed orders from first-time visitors, measured by revenue per visitor."

Kite writes the offer copy for each version using what it knows about your products and your brand voice, so the page explains the offer plainly before the shopper reaches checkout. For ideas beyond these, the guide Cross-sell examples at checkout that raised order value, and how to test your own Coming soon collects offers other stores have tried.

Kite gives every version of the test its own checkout, so the payments for each offer stay separate. In your connected Stripe account, it creates what the offer needs. For a discount, that's a coupon and, if shoppers should type a code, a promotion code. For a bundle, it's a product and a price for the bundle. For free shipping, it's a shipping rate set to zero.

Then Kite creates a payment link for each version. A payment link is a ready-made Stripe checkout page with its own web address. The link for your current offer stays as it is, and the link for the new version carries the offer: the bundle price, the add-on as a second item, the free shipping rate, or the option to enter a promotion code. Each version of your page sends its buy button to its own link.

Kite can also limit a test code so it doesn't spread beyond the test. A Stripe promotion code can expire on a set date, stop after a set number of uses, or require a minimum order. That keeps a code shared on a coupon site from turning a test into a store-wide discount.

For an offer after payment, Kite sets the main payment link to send buyers to a confirmation page on your website when they finish paying. That page presents the related product with its own payment link.

An A/B test shows your current page and checkout to some visitors and the version with the new offer to others, at the same time, then compares the results. On a website Kite hosts, Kite sets up the test. It chooses the conversion to measure from the events your site already records, such as clicks on the buy button. Then it builds the new version of the page in its own draft, with the offer copy and the new checkout link, and splits traffic between the two versions, half and half by default.

Kite shows you the test before it runs. You see the idea being tested, the conversion it measures, the traffic split, and the Stripe offer behind the new checkout, and you can preview the new version. When you're ready, you start the test from your website's Experiments panel in Kite. The plan locks when the test starts, so the measured conversion and the traffic split stay the same for the whole run.

While the test runs, Kite tracks how many visitors saw each version, how many clicked through to checkout, and the chance that each version is better. It waits for the planned minimum number of days and visitors before it recommends anything. If the difference is still too small to trust after three times that minimum, Kite recommends ending the test as inconclusive and keeping your current offer.

If your website's code lives in GitHub, you can import the repository into Kite, which lets Kite run the same kind of test on it.

A checkout offer test is finished when you know which version earned more money for every shopper who saw it. Clicks on the buy button tell you which offer was more tempting. They don't tell you which one paid off.

So Kite follows each version into Stripe. Because every version has its own payment link, Kite counts the completed checkouts for each one and adds up what those orders were worth after the discount. It divides that revenue by the number of visitors who saw the version. That figure is revenue per visitor, and it captures both halves of the result: how many people bought, and how much they spent.

Here's why that matters. Suppose your current checkout turns 3% of visitors into buyers at an average order of $60, which is $1.80 per visitor. A version with 15% off turns 3.4% into buyers at an average of $51, which is about $1.73 per visitor. The discount won more orders and still lost money. A free shipping offer above $75 might convert slightly fewer shoppers but lift the average order to $78, and come out ahead of both.

Kite also reads the abandoned side. It compares how many checkouts expired unpaid for each version, which shows whether an offer removed a reason to leave or only changed who bought. Free shipping still costs you the postage, so when a shipping offer wins on revenue, Kite asks what your shipping actually costs you and weighs that against the extra orders before it recommends keeping the offer.

Every test ends with a verdict: the new offer won, it lost, or the result was inconclusive. Kite reports the numbers for each version, how long the test ran, and anything that might have affected it, such as a sale or a campaign that changed who was visiting. If the new offer won, Kite asks you to approve making it permanent, then merges the change. It records the test and its lesson in your team's knowledge base, so later tests build on it, and proposes the next offer to try.

If you're also wondering whether your prices themselves are right, Kite can test your pricing and pricing page with the same approach.