How to run a CRO audit on your own store
A useful CRO audit is mostly arithmetic rather than opinion. You're looking for the step where the most people with real intent stop, and you can find it with the analytics you already have. This is the method we use, in the order we use it.
Most audits go wrong because they start at the product page. That's where it's easiest to have opinions, and it's usually not where the leverage is.
What do I measure before looking at anything?
Four numbers, split by device. Blended figures hide the problem, because mobile and desktop behave so differently they might as well be separate businesses. Our post on why Shopify traffic doesn't convert shows how big that gap gets.
- Sessions, and where they came from
- Sessions that viewed a product
- Sessions that added to cart
- Sessions that reached checkout, and sessions that completed it
Write them down as a funnel, with the drop between each step as a percentage. One bad-looking number on its own doesn't tell you much. What you want is the largest single drop, because that's where the work is.
Where is the drop usually?
At checkout, more often than most store owners expect. Baymard Institute puts average cart abandonment at 70.22% across fifty studies, and mobile is worse: 80.02% of mobile carts are abandoned, compared with 66.41% on desktop.
Baymard estimates that better checkout design can bring a 35.26% increase in conversion rate for a large site. That's a big number, and it comes after everything most audits spend their time on.
What causes checkout abandonment?
These are Baymard's ranked reasons, and the order matters, because the first one is almost double the second:
- Extra costs too high (shipping, taxes, fees): 48%
- Forced account creation: 26%
- Checkout too long or complicated: 22%
- Didn't trust the site with card details: 18%
- Total cost not shown upfront: 17%
The top reason isn't really about price. It's about surprise: a number that showed up later than the shopper expected. A store that shows clear shipping costs on the product page can beat a cheaper competitor that only reveals them at step three.
On the other hand, around 42% of people who abandon say they were just browsing and weren't ready to buy. You can't win those back, and any audit that promises to is selling you something.
How do I audit my own checkout?
Buy something. For real, with your own card, on your own phone, on cellular data instead of office wifi. Then do it again as a new customer, using an address that doesn't autofill.
Count the steps and the form fields. Note the exact moment you first see the shipping cost, and whether anything about it catches you off guard. Note whether you were asked to create an account, and whether guest checkout was easy to find or tucked away.
Most of what a paid audit would tell you about your checkout, you'll find in those ten minutes.
What should I check on mobile specifically?
Start with load time, measured on a real phone rather than a desktop emulator. A Google and Deloitte study found that a 0.1-second improvement in mobile speed raised conversion by 8.4% and average order value by 9.2%, so this is more than a box to check.
Then check the mechanical stuff: whether buttons are big enough to tap while you're walking, whether the ZIP code field brings up a number keyboard, whether a sticky element covers the buy button, and whether a pop-up fires before the page is usable.
What should I deliberately ignore?
Button color. Hero copy rewrites. Anything a tool flagged as a "best practice" without looking at your funnel. These eat up an audit's time and almost never turn out to be the biggest drop.
Also skip any test you can't run to significance. If you get 400 orders a month, you don't have the traffic to detect a 5% lift, and running the test anyway just gives you a number that feels like evidence but isn't.
What does the output look like?
A short list, ordered by expected value rather than by how easy each fix is. For each item, note the drop it addresses, how many sessions it affects per month, and what you'd have to change. If you can't tie an item to a step in the funnel you measured, it doesn't belong on the list.
Three items done properly will beat twenty observations. Measuring first is what lets you confidently ignore most of the things you could be doing.
This method is where our own CRO work starts. At Frog Fuel, that work runs alongside Shopify Plus, Klaviyo and Skio as part of a long-term partnership. If you'd rather have someone else run it, we do this as a free audit. But the method above is the whole method, and there's nothing stopping you from doing it yourself this afternoon.