Checkout friction audit
Use when checkout abandonment is high and you need to locate where and why people drop.
Fill in before running
Replace each placeholder with your own detail. The more specific you are, the less the model invents.
- {{STEP_LIST}}
- {{STEP_DROPOFF_DATA}}
- {{PAYMENT_METHODS}}
- {{DELIVERY_MODEL}}
- {{GUEST_OR_ACCOUNT}}
Getting a better result
- Include step level drop-off figures if you have them; without them every finding is a heuristic guess.
- List the error messages verbatim, since bad error copy is often the largest single leak.
- Ask it separately to run the same audit assuming a returning customer.
Questions about this prompt
When do I audit the flow rather than the form on one screen?
When abandonment is high across a multi step checkout and you cannot say which step leaks. The field reduction prompt looks inside a single form; this walks every step, including where the total price becomes known and any point that sends someone to an inbox or another app. Run it before redesigning any one screen.
What do I need in front of me before running it?
The steps in order, drop-off by step if you have it, the payment methods, the delivery model and whether guest checkout exists. Without step level figures every row comes back labelled heuristic. Paste the error messages exactly as they appear, since bad error copy is often the single largest leak in a checkout.
What comes back, and which part is worth acting on?
A row per step labelled data or heuristic, four targeted answers, and fixes ranked by expected impact over effort with one named as first to ship. The answer to read first is where the total price including delivery and tax becomes fully known, because a late reveal explains more abandonment than most interface problems.
What is the mistake that costs me here?
Auditing only the first time buyer. A returning customer with a saved card meets a different flow, so run the audit again on that assumption and compare. Take the note about one-click payment methods seriously as well: the prompt names the integration work rather than treating them as a quick win, and that estimate is usually right.