You don’t need a CRO consultant for the first lap of checkout improvements. You need a burner phone, a stopwatch, and a couple of hours.
Here’s the audit I run myself before I ever quote a client. They could run it themselves — most don’t, which is the whole point of this note.
What you need. A cheap Android phone, ideally one with a small screen and weak processor — the closer to the median customer’s device, the better. A stopwatch. Notes app open. A real credit card you’re willing to charge $1 to and refund.
The audit.
1. Open the homepage. Time how long until the LCP image appears. Anything over 2.5 seconds is a problem.
2. Tap the first product card you see. Time from tap to a usable PDP. If you wait 1.2 seconds with nothing changing, that’s where customers bounce.
3. On the PDP, count the taps to add-to-cart with the most common variant combination. More than three taps for a single-variant product is too many.
4. Add to cart. Watch carefully — does the cart drawer slide in immediately, or do you stare at a blank screen for a second?
5. Open the cart. Count the upsells. Three or more is too many. They’re actively hurting conversion at this stage.
6. Tap checkout. Time from tap to the email field being focused and usable. Over 1.5 seconds means the checkout is loading too much JS.
7. Try to type your email. Does the keyboard cover the field? Does the page scroll automatically to keep the field in view? If not, you’re losing thumbs that don’t want to fight.
8. Enter a deliberately invalid postcode. Does the error appear inline, or only after you tap continue? Inline saves you a tap and a frustration spike.
9. On the payment step, watch for autofill behavior. Does the card field accept paste? Does Apple Pay show up immediately, or does it lag in?
10. Complete the purchase. Note every place you paused, every moment of friction, every time you wondered “did that work?”
11. Read the order confirmation email. Is it useful or is it a generic Shopify default with your logo glued on top?
That’s the audit. It takes about two hours including write-up. The list of fixes that drops out of it is almost always longer than the fixes any consultant would propose in the first $5,000 of an engagement.
The reason most founders don’t do it is that doing it once means you have to do it again every quarter as the site changes. That’s also the reason it works.