Every ecommerce audit we run starts the same way, with a rule we apply to ourselves before anyone touches a theme file, a feed, or a bid. Change nothing for two weeks. The temptation on a new store is to fix the obvious things immediately, and the obvious things are usually cosmetic. If you start editing before you have measured, you lose the only clean read you will ever get of how the store behaved before you arrived, and you will spend the next quarter unable to prove which of your changes did anything.
The second rule is that an audit is a ranked list, not a document. A hundred findings with no order is a way of making the client responsible for prioritisation, which is the part they were paying us to do. We finish with somewhere between eight and fifteen items, each tied to a number the store already reports, sorted by expected revenue impact divided by effort. Everything else goes in an appendix that nobody has to read.
What An Ecommerce Audit Covers Before The Fun Parts
Tracking comes first, always. If the measurement layer is wrong, every other finding in the ecommerce audit is built on sand. We check that purchase events fire once and only once, that values are sent in the correct currency including or excluding tax consistently, that refunds are reflected somewhere, and that the number in the analytics platform reconciles with the number in the store’s own order table. A gap of more than a few percent between those two is common in stores that have never been audited, and until it is closed, no channel comparison is meaningful.
Then inventory of the catalogue. How many products are published, in stock, carrying a unique description, carrying more than one image, or sitting in a category no navigation link reaches. Stores that grew through a migration usually carry a long tail of orphaned products, indexed and crawled but unreachable by humans.
The Money Path
We walk the full purchase path on a real device with a real payment method, not a staging environment, and we do it on mobile first because that is where most of the traffic and most of the abandonment lives. Every field, every validation message, every shipping estimate, every unexpected cost.
The findings here are dull and worth more than anything else in the report. Shipping cost revealed only at the last step. A required account creation that has no business reason to exist. A discount code field that sends people off to search for a code they will not find. Address validation that rejects legitimate local formats. In the stores we work with, checkout friction of this kind accounts for a larger share of lost revenue than the ad account does, and it is cheaper to fix.
Speed, Measured Where Customers Are
Run the measurement on a mid range phone on a normal mobile connection, not on a desktop with fibre. The number that matters is time until the main product image and the add to cart control are usable, not a composite score. Then find out what is actually costing the time, which is almost never the platform itself.
- Third party scripts: chat widgets, heat mapping, review platforms, and abandoned tag manager containers that nobody has pruned in years
- Uncompressed hero and product images shipped at desktop dimensions to phones
- App or plugin bloat, where installed tools load their assets on every page rather than the one page they serve
- Fonts loaded in four weights when the design uses two
- Sliders and carousels on the home page, which cost load time and, in the tests we have run, rarely earn it back
We keep a before and after measurement for each of these so the improvement is attributable. Removing one abandoned tracking script is a five minute job that sometimes returns half a second, and half a second on a product page is worth more than most redesigns.
Merchandising And Search
Pull the internal site search queries for the last ninety days and read them properly. This is the cheapest source of truth in any ecommerce audit, because it is customers telling you in their own words what they came for and what they could not find. Queries with high volume and zero results are either a catalogue gap or a naming mismatch, and both are fixable in an afternoon.
Next, the category and collection structure. Are the pages that carry commercial search demand actual landing pages with copy, or are they empty filtered views? Is the sort order the store’s default, which is usually alphabetical or by date added, rather than by margin or conversion rate? Do out of stock products sit at the top of collection pages? Default sort order is one of the highest leverage changes available on most stores and it requires no design work.
Acquisition, Last
We deliberately look at the ad accounts at the end, once we know what the store can convert and what it cannot. Feed quality first: titles that lead with the attributes people search, product types mapped, GTINs present, and no products disapproved for reasons nobody has read in months. Then structure, then budget, then creative.
The output is a sequenced plan rather than a list of complaints. Fix tracking, then checkout, then the worst speed offenders, then merchandising, then spend. Each item names the metric it should move and the window in which it should move. An ecommerce audit that ends without that sequencing is just an opinion with a table of contents, and we have all read enough of those.