Test a membership upsell modal without breaking checkout

The checks that matter

Most upsell tests stop when the dialog appears. The failure that costs conversions happens after that, when the shopper says no.

The last row is the assertion worth copying. A modal that closes but leaves the page behind it in a broken or scroll-locked state is a common defect and it passes any test that only checks the dialog disappeared.

The same flow as a test file

The trigger is a paid express slot rather than a button, because that is how the upsell actually fires. Testing it through its real trigger is what makes the result meaningful.

---
mode: testing
url: https://my-testing-repo-main.vercel.app/walmart-clone-app?reset=true
max_steps: 45
tags: [walmart, membership, modal]
---

# Wallmark 4.3: Membership upsell

Catalog objective: trigger the Wallmark+ banner during checkout and open the signup modal.
Key assertion: plan pricing renders and the modal closes cleanly.

## Add an item and open checkout
Click "Add to cart" on "Large Eggs, 12 ct", click "Continue to checkout", and verify the checkout page shows a "Pickup time" section.

## Trigger the upsell with a paid express slot
Click "Tomorrow, 12pm - 1pm · Express" and verify a dialog appears titled "Save on this order with Wallmark+".

## Confirm plan pricing renders
Verify the dialog shows "Annual - $98.00/year" and "Monthly - $12.95/month".

## Close the modal cleanly
Click "No thanks" and verify the dialog is no longer shown and the checkout page still shows the "Order summary" with a "Place order" button.

Both plan prices are quoted in full, including the period. A modal showing the annual price against a monthly label is a pricing error, and only an exact string assertion catches it. Run the test with:

kane-cli testmd run tests/walmart/membership-upsell_test.md

Run output

Four passes in 158 seconds. Confirming the plan pricing was the cheapest step at 30 seconds, since the dialog was already open by then.

---
test: ../membership-upsell_test.md
status: passed
started: 2026-09-15T09:59:50.856Z
duration_s: 158
session_id: b208f48d-4744-4992-97d1-80b870947ead
---

# Wallmark 4.3: Membership upsell - Result

## Add an item and open checkout ✓ passed (43.5s)
md5: 9af6b6dbab7427b8cc54bc1d8096b985
Click "Add to cart" on "Large Eggs, 12 ct", click "Continue to checkout", and verify the checkout page shows a "Pickup time" section.

## Trigger the upsell with a paid express slot ✓ passed (40s)
md5: 9705338a67f954e80531ec6413f428b0
Click "Tomorrow, 12pm - 1pm · Express" and verify a dialog appears titled "Save on this order with Wallmark+".

## Confirm plan pricing renders ✓ passed (30s)
md5: e6732dac46d43a1831e0bcf9783aab02
Verify the dialog shows "Annual - $98.00/year" and "Monthly - $12.95/month".

## Close the modal cleanly ✓ passed (40.5s)
md5: 0638cfee5acba716d26f5f17850a6178
Click "No thanks" and verify the dialog is no longer shown and the checkout page still shows the "Order summary" with a "Place order" button.

Closing the modal cost 40.5 seconds, more than opening it, because that step asserts three separate things: the dialog is gone, the order summary is back, and the Place order button is present.

Taking it further

You end up with a conversion-path test that covers the decline as carefully as the accept. That is the right emphasis, since most shoppers decline and all of them need a working checkout afterwards.

Upsell copy and pricing change often, which makes this a good candidate for running on every release. Kane CLI in a pipeline covers scheduling that, and checking UI before a pull request covers catching it earlier.