Methodology

How UXFix audits an online store

Every UXFix audit, the free ones and the published case studies alike, is produced the same way: a scripted purchase walk over the store's public pages, judged against a fixed rulebook, scored by a published formula. This page is the whole method. Nothing on any audit page is invented, and everything on one can be traced back to a rule described here.

The walk

The audit opens the store the way a shopper does: it finds a real product page, reads it, adds the product to the cart, opens the cart, and proceeds through checkout to the payment step. It runs on desktop and mobile viewports and captures a screenshot of every screen it judges, which is why each published finding can show the screen it is about.

It sees only what the public can see. No analytics access, no store login, no cooperation from the store. If a rule cannot be judged from the public storefront it is dropped from the score and reported as not assessable, never guessed at.

The rulebook

The rulebook currently holds 115 deterministic rules: 40 on the product page, 25 on the cart and 50 on the checkout, because that is the order abandonment research says the money is lost in. 103 rules judge the experience a human shopper gets; 12 judge whether an AI shopping agent can read price, stock and shipping and complete a purchase, reported separately as the AI Shoppers score.

Each rule has a severity (critical, high, medium or low) and produces one of five results: pass, partial, fail, not assessable, or not applicable. A fail is a positive claim: it means we checked and the store got it wrong. When a detector is not certain, it must answer not assessable, and that rule leaves the score entirely.

The score

Every rule's weight is computed from exactly two inputs: its severity and its page's share of the score. The checkout carries half the score, the product page thirty percent and the cart twenty, matching where abandonment actually happens. A pass earns the full weight, a partial earns half, a fail earns nothing. The score is the weighted share of judged rules the store gets right, from 0 to 100.

The score is always reported next to its coverage: how much of the applicable rulebook we could actually judge on that store on that day. A score without its coverage is not an honest claim, which is why every UXFix surface prints them together.

What the conversion-loss estimates mean

Each failing rule carries an estimated conversion loss, shown as a range. These are modelled from published cart and checkout abandonment research, applied per severity and discounted for partial results. They are estimates, clearly labelled as such, never measurements from the store's own analytics, which we do not have. When a store shares its real traffic numbers, the same model produces dollar ranges instead of percentages.

What a published case study may and may not say

A published audit names a real business, so its page obeys rules the private reports do not need. Every number on it is read from the saved report. A finding whose wording does not read as a sentence a person would write is withheld, and the page says how many were withheld. A screenshot we cannot confirm belongs to the step being judged is withheld too, with the verdict kept. Where the page compares a store to others, the medians and ranks are computed from the full set of published audits, the same set the public gallery lists.

Consent: audits we run of well-known brands are commentary on public web pages and are published by our staff. A customer's audit of their own store is private unless the customer agrees to publication, and that consent is recorded. Anyone at a named store can have their page corrected or removed by emailing [email protected].

Freshness

Every audit page states the date its crawl ran, and its findings describe the store as captured on that date. Stores change. When we re-run and republish an audit the page updates in place at the same address, and anyone can ask for a re-check through the email above.

Independence

No published audit is sponsored, commissioned, paid for or reviewed by the store it describes, and UXFix has no commercial relationship with any store it publishes commentary on.