A Published Audit

Forty-three findings on one storefront.

This is a real store audit, redacted. The business is not named. Five of its findings and one open item are set out below exactly as they were written. If you are considering paying us to audit your store, you should be able to read our work first.

43Findings
12Categories rated
5Published in full
1Open item logged
The store01 / 05

A direct-to-consumer fashion label, three years trading.

Founded and run from outside the market it sells into. The storefront layer had not been materially changed in fifteen months, which is most of why it had drifted.

The subject, anonymised
BusinessDirect-to-consumer fashion label
TradingThree years
PlatformShopify
Market and order valueMainly the United States, around $60

The audit was run across twelve categories and produced forty-three findings. Five of them are set out below in full, along with one item we deliberately did not record as a finding at all. The rest were graded medium and low, and are the ordinary housekeeping any three-year-old store accumulates.

Six of the forty-three02 / 05

Every finding names a location, an observation, and what it meant.

“Your images are too large” is not a finding. Which images, on which page, at what size, and what it costs: that is a finding. The same discipline applies to the legal and configuration items below.

Finding 01CriticalMarket and currency configuration

The market configuration nobody had checked

Three regional storefronts were configured. The founders’ own market was not one of them, so every visitor from their home country fell through to the default: dollar pricing, and a banner promising free delivery across the United States, while the published shipping policy specifically committed to delivering to that country in 9 to 11 business days.

Why it matters

Their own countrymen were being shown a storefront built for somewhere else, on a site that had promised in writing to ship to them.

Finding 02Open item, not a findingMarket and currency configuration

A currency stack that looked fine and might not have been

A third-party app converted displayed prices to each visitor’s local currency in the browser. Underneath, the store’s configured market currency was unchanged. So a visitor saw pounds while the configuration said dollars.

Whether the checkout then charged in pounds, or reverted to dollars with a foreign exchange fee attached, is the difference between a clean sale and a chargeback. We could not establish that without placing a test order, so we did not claim it.

The two-minute test we gave the owner

Add to cart from the market in question, go through to the payment step, and read the currency.

Why it stayed an open item

It went into the report as an open item rather than a finding, with the test attached. Every report we write carries a section listing what we could not verify and why.

Finding 03CriticalReviews and trust signals

A star rating going to Google that no customer had given

The review app was carrying seed data, and an aggregate rating derived from it was syndicating into Google Shopping.

Seed data is a default in more than one popular review app, not a decision anyone made. It is the kind of thing that is genuinely easy to miss.

Why it matters

Since April 2025, UK rules on how businesses handle reviews have had a regulator behind them that no longer needs a court to act. This is not a cosmetic finding.

Finding 04HighLegal documents and disclosures

Contract documents pointing at a dead domain

The terms customers were agreeing to at checkout linked out to a domain that no longer resolved.

Why it matters

The agreement was live. Part of what it referred to was not.

Finding 05HighConsumer law and distance selling

Terms and refunds contradicting each other

The terms nominated a governing law under which the store’s own published refund policy did not work.

Why it matters

Two documents, both live, each describing a different deal.

Finding 06MediumApps, integrations and cost

A paid app that could not do its job

A merchandising engine was installed, subscribed and running against product data that did not contain the fields it needed to function.

Why it matters

A monthly bill for a capability that was structurally unable to operate.

The twelve03 / 05

What was rated.

Each category is scored in the report, with the method stated so the numbers can be checked, and it closes on a ranked priority list: what to fix first, judged on impact against effort. That is the section the report is built to end on. The five marked below are the ones the published findings came from.

  • 01Market and currency configurationPublished
  • 02Product and collection pages
  • 03Cart and checkout
  • 04Product data and structured markup
  • 05Performance and Core Web Vitals
  • 06Findability and indexation
  • 07Accessibility
  • 08Security and headers
  • 09Legal documents and disclosuresPublished
  • 10Consumer law and distance sellingPublished
  • 11Reviews and trust signalsPublished
  • 12Apps, integrations and costPublished
On method04 / 05

The audit was wrong twice before it was right.

That report went through three versions before it was fit to send. We would rather tell you that than not. An audit nobody has reviewed against itself is an opinion with a scorecard attached.

  • Version one

    The first pass, written up and then reviewed against itself rather than sent.

  • Version two

    Corrected four findings from the first, including one recommendation that would have pushed a wrong measurement into production if anyone had acted on it.

  • Version three

    Corrected the framing again, once the trading timeline was properly established. That is the version the client received.

Finding 02 above is the same discipline in public. We could see the configuration. We could not prove what the checkout actually charged. So it went in as an open question with the test attached, not as a conclusion. When a finding does not survive review it comes out, and the correction is recorded.

On naming05 / 05

Why the business is not named.

Audit clients are anonymised by default. We publish findings, not names. Nobody wants their defects public, and discretion is part of what you are paying for. If a client would rather be named, that is their decision to make and not ours.

Our website builds are named, because a finished site is something a client is pleased with. An audit is the opposite kind of document. The two are treated differently on purpose, and you can see both on this site: the named build work is here.

If you want oneTwo routes
Free

Mini audit

Your homepage and one page you nominate. Five scores and five findings, each one checkable by you in ten minutes on your own site. Two business days, no call, no obligation.

Request the mini audit
£899, total payable

E-commerce Store Audit

The audit on this page. Twelve sections, up to eight pages plus a complete checkout run-through on a real device, a written report and a 45-minute debrief. Five working days.

See what it covers

Want to know what is on your own store?

Start with the free mini audit. Five findings, two business days, and you can verify every one of them yourself.

Request