For ecommerce founders and marketing teams deciding whether their store needs focused fixes, a redesign or better measurement.
Updated
Start with a decision, not a score
Choose the question the audit needs to answer. For example: are mobile visitors unable to choose a variant, or do they leave after discovering delivery costs? A useful audit links a specific observation to a proposed change and a way to check it.
Use one issue record: affected page and device; evidence; likely customer obstacle; proposed change; owner; effort; acceptance check. This makes a backlog reviewable without pretending that every design preference is a conversion problem.
Check that the funnel describes real behaviour
Confirm how product views, add-to-cart, checkout starts and completed orders are measured. Check duplicate events and compare reported orders with the trading system before interpreting a drop-off. Use aggregated reports and access controls rather than exporting customer details into an audit document.
Review mobile and desktop separately, then look at landing pages and traffic sources. A campaign sending visitors to the wrong product can look like a page problem. Note promotions, stock-outs and delivery changes alongside the measurement window.
Walk the mobile journey with a real buying task
Choose a product, select an available variant, find delivery and return information, add it to the cart and begin checkout. Check narrow screens, long product names, validation errors and slow loading. Test with realistic catalogue content rather than a perfect sample product.
- Product choice
- Can a customer distinguish size, material, quantity and availability before adding the item?
- Product confidence
- Do images and copy explain scale, included items and limitations accurately?
- Delivery and payment
- Are relevant charges and delivery information available before the final commitment? Confirm payment methods against the configured provider.
- Cart continuity
- Do quantity changes, discounts and validation messages preserve the customer’s choices?
Separate confirmed faults from hypotheses
A selector that fails on a supported device is a reproducible fault. Moving a product image higher is a hypothesis unless evidence supports the diagnosis. Fix broken purchase paths first; then weigh likely commercial relevance, evidence strength, implementation effort and reversibility.
Do not label an attractive redesign as a proven uplift. Keep the original observation and the final implementation together so the team can later assess whether it addressed the problem.
Choose a measurement method the traffic can support
Use a controlled experiment when the available traffic, tooling and business constraints make it suitable. Agree the primary measure, guardrails and analysis window before starting. Avoid ending a test only because an early result looks favourable.
For smaller stores, validate that the fault is gone and review subsequent trading carefully. A before-and-after comparison can inform the next decision, but promotions, audience changes and seasonality prevent it from proving that the website change caused the result.
What a useful audit handover contains
Ask for a short diagnosis, supporting evidence, a prioritised backlog and acceptance criteria for each recommended change. The next step should be a clear implementation brief, a request for better data or a decision that a broader redesign is warranted.
You do not need a large checklist to start a conversation. Share your store URL, the journey causing concern and the data you currently trust.