Audit what works before what doesn't. A redesign changes two hundred things at once: if you don't know which ones were earning money, you won't know what you broke.
The authors know the temptation from the inside. When their own agency kicked off a redesign of its website, the memo that opened the project already hedged:
"it's not productive for an art director to have 24 people commenting on a mockup."
On an online store, the audit plays that safeguard role: it puts the numbers on the table before opinions settle in.
The benchmark
The audit takes the form of an inventory in three columns (keep, fix, rebuild). It covers the pages that drive revenue, the URLs that receive traffic, the features your customers actually use, the apps and the tracking setup. It comes with a baseline measured over eight clean weeks of the current site.
This year, those eight weeks create a calendar problem. Shopify changed how it counts sessions between September 21 and 23, 2026, and its help page says to use the later data as your new basis for comparison.
Counting from September 24, eight weeks end on November 18, nine days before Black Friday on November 27. If your new site is due to launch in January or February, this is your only clean window, and it's already open.
The rule
Don't launch a redesign without a baseline, or without a written list of what must not get worse. Copy this, then rewrite it with your own numbers:
Period measured: September 24 to November 18, 2026, with sessions counted under Shopify's September 2026 rule.
1. The add-to-cart rate of your five top-earning product pages.
2. The revenue from pages that receive search traffic.
3. The mobile speed of the product page and collection templates.
Review two weeks after launch, then again at six weeks.
This document fits on one screen. It also gives you the line to say to your leadership before the first workshop:
"We know what the current site earns and where; the redesign will be judged on these three lines."
The steps
- Templates. Rank them by revenue and by conversion rate.
- URLs. List the pages that capture search traffic, then look for the ones no search tool sees: the QR codes printed on your packaging, the links in your automated emails, the links from the creators you work with, the landing pages of your ads. These links break silently, and Shopify only redirects an old URL if it returns a 404 error.
- Features. List the ones your customers use: search, filters, one-click checkout, back-in-stock alerts.
- Apps and scripts. Inventory them with the reason each exists, and note who installed them.
- The starting point. Pull your numbers over the window described above, before November promotions distort them.
The trap
The most common one is redoing the site because it's four years old, a criterion no report measures. Decided on taste, a redesign often swaps a page that converted for a page the committee likes.
The other trap is launch day. While discussing the terms of a project, a client told the authors:
"Of course the live site has to be perfect."
They wanted the same thing she did, and they also knew there are almost always fixes in the weeks that follow.
Budget for that period, and keep your list close at hand so you know what to fix first.
