The Shopify store audit checklist we actually use

Why most store audits are worthless

The typical free store audit is a PageSpeed screenshot, a note that your images are large, a remark about trust badges, and a call to action. It is a lead magnet wearing the costume of a diagnosis.

A useful audit does two things that one does not. It measures against a benchmark, so "your conversion rate is low" becomes a number with a distribution behind it instead of an adjective. And it ranks what it finds by what the fix is worth, because a list of eleven problems with no order is a list you will not action.

What follows is the structure we use on a paid teardown, in the order we use it, with the tools for each section. You can run it yourself. It will cost you a focused afternoon.

01 — Write the summary last, put it first

Leave a blank page at the top and come back to it.

When you have finished, write one paragraph: what this store is leaving on the table, and the single highest-impact fix. If you cannot compress it to a paragraph, you have a list of observations rather than a diagnosis, and going back through the sections to find the through-line is the actual work.

02 — Speed, on both viewports, against thresholds

Run PageSpeed Insights on your homepage and on your best-selling product page. Both mobile and desktop. The product page matters more and gets skipped more.

Record four numbers. Three of them have a published Google threshold:

For the overall PageSpeed score, Google's own bands are 0 to 49 poor, 50 to 89 needs improvement, 90 to 100 good. We hold stores to 80 or above, which is our bar rather than Google's — worth knowing when someone quotes a target at you without saying whose it is.

Then find the cause, not the score. Open DevTools, load the product page, sort the network panel by size, and write down the three heaviest things and what they weigh. This is the difference between a finding and a complaint: "the hero image is a 2.1MB PNG" is actionable, "images could be optimised" is not.

Two things account for most Shopify slowness in practice: uncompressed hero imagery, and apps. Every app that injects a script costs you something on every page load, including the ones you stopped using. Check your theme for leftover script tags from apps you have uninstalled.

03 — The mobile buying path, walked properly

Set a browser to 390 by 844 — an iPhone-class viewport — and walk from the homepage to a filled cart. Do not click through as the person who built it. Walk it as someone who has never seen it.

Three questions, and be strict:

• Is the add to cart visible without scrolling on the product page? On most themes, on most product pages, it is not.
• Does it stay reachable once you scroll? Scroll to the reviews and try to buy. If you have to scroll back up, that is a sticky add-to-cart you do not have.
• Do the trust signals appear before the add to cart, or after it? Shipping cost, returns window, delivery estimate. After the button means after the decision.

Count the taps from homepage to cart. Write down where you hesitated, even for half a second. Especially that.

04 — The desktop path, reported separately

Same journey at 1440 by 900.

Report it separately rather than merging the two, because a store can pass on desktop and fail on the device most of its traffic uses — and if you only look at the merged picture you will underweight the failure. Check your own analytics for the mobile share of sessions before you decide which one to fix first.

05 — AI visibility, the section nobody else runs

This is the one that is missing from every free audit and most paid ones.

Write the three questions a real buyer would type before they had heard of you — the problem, not your brand name. Ask each in ChatGPT, Gemini and Perplexity. Record which stores get named and which sources the answers lean on.

Then check the mechanics:

• Open yourstore.com/robots.txt and read which AI crawlers you have blocked, deliberately or otherwise.
• Run a product URL through Google's Rich Results Test. You want a valid Product node with an Offer carrying price, currency and availability — and exactly one of them. Two apps each injecting a Product node is common and damaging.
• Read the first hundred words of your best product page and ask whether they state what it is, who it is for, and what it costs, in extractable text rather than in an image or a click-to-open tab.

There is more on the method in how to get your store recommended by ChatGPT.

06 — Now rank what you found

This is where an audit becomes useful, and where nearly all of them stop.

Take everything you wrote down and rank it by what the fix is worth, not by how easy it is or how much it annoys you. For each item, write three things: the problem with its measurement, the specific change, and the estimated impact. Then keep the top five and delete the rest.

Deleting the rest is not a formatting choice. A list of eleven problems does not get actioned; five ranked ones do.

Each surviving item should be specific enough to hand to whoever will do it. "Compress the hero image from a 2.1MB PNG to WebP under 200KB" is a fix. "Improve images" is a category.

07 — Put a number on it, conservatively

You need a figure or the ranking is just an opinion with an order.

The method we use: take monthly ad spend and treat 3% as the conservative recovery estimate and 7% as the aggressive one. Lead with the conservative number, always. The aggressive figure makes a better slide and is the number you will be held to in three months, which is a bad trade.

Then split that conservative envelope across the five fixes rather than estimating each one separately and adding them up. If you estimate independently, your five fixes will sum to a different number than your headline, and the first thing a sceptical reader does is add up the five.

For framing, our own working benchmark set — reviewed quarterly, and ours rather than a published industry source — puts the median site conversion rate at 1.17% with most stores between 0.92% and 1.52%, and email and SMS at 9 to 15% of revenue for most stores against 25 to 40% once it is genuinely built. That last row is the widest gap in most audits and the easiest to act on.

08 — Decide what happens next, and be willing to say nothing

Write down one recommendation. Not a menu.

If the honest recommendation is that you fix two of these yourself and buy nothing, that is a legitimate outcome of an audit and the most trustworthy one available. An audit that always concludes with "and therefore hire us" was a sales document with a measurement section attached.

09 — The part you cannot do for yourself

Two things, honestly.

The first is a second pair of eyes. You cannot walk your own checkout as a stranger, because you know where everything is. Whatever you find, someone who has never seen the store will find something different, and usually something worse.

The second is the ranking. Measuring is the easy half; deciding that fix three is worth more than fix one, and why, is the part that takes having done it on a lot of stores. This checklist will get you the findings. Sequencing them well is a different skill.

If you would rather have it done, ranked and written up, that is the Store Teardown: €750, five days, nine sections in this order, plus a recorded walkthrough, and no call in the process. Whatever you decide afterwards, the document is yours.