Start with the business question

“Make the site faster” is too broad to guide an audit. Start by naming the customer journey and the commercial or operational effect. Product pages may feel slow on mobile. Search may hold up category browsing. Checkout may hesitate after a payment or delivery request. Each symptom points to a different investigation.

Define the affected devices, templates, markets and trading periods. Record what success would change: a faster interaction, fewer errors, more completed checkouts or a release process the team can use confidently.

Use field data before lab scores

Lab tools are valuable because they are repeatable, but they represent a controlled visit. Real-user monitoring and Core Web Vitals show what customers experienced across devices, networks and sessions. Use field data to find the affected population, then use lab traces to reproduce and explain it.

Look beyond one headline score

Largest Contentful Paint, Interaction to Next Paint and Cumulative Layout Shift describe different parts of the experience. A page can render its main content quickly while still responding slowly to filters or adding products to the basket. Treat each measure as evidence about a journey, not a grade for the whole platform.

Test the journeys that matter

An audit should cover representative category, search, product and checkout paths. Include a cold visit, a returning visit and the devices customers actually use. Test logged-in or personalised states where they change rendering or API behaviour.

  • Record server response, critical rendering and main-thread work separately.
  • Identify the image, font, script or request associated with the slowest visible step.
  • Repeat tests to distinguish a stable constraint from an isolated slow request.
  • Inspect third-party behaviour after consent as well as before it.

Separate frontend, platform and third-party causes

Magento performance problems often cross layers. A large hero image is visible in the browser, but an uncached page or slow product query may have delayed the request before it. A filter interaction may be blocked by frontend JavaScript, the search service or both.

Frontend delivery

Review image formats and sizing, font loading, JavaScript execution, CSS delivery, layout stability and the order in which critical assets become available.

Magento and infrastructure

Inspect full-page cache behaviour, cache invalidation, application and database traces, indexers, cron health, extension overhead and the path between the edge, web tier and downstream services.

Third parties

Tag managers, reviews, personalisation, search, payments and marketing tools can all compete for network and main-thread time. Measure their contribution and confirm which features the business genuinely needs on each template.

Prioritise by impact, confidence and dependency

A long list of findings is not yet a roadmap. Group work by the customer journey affected, the strength of the evidence, expected impact, delivery effort and dependency on other teams or suppliers.

Quick fixes are useful when they remove a proven constraint. They are less useful when they postpone an architectural decision the team already needs to make. Make that trade-off explicit.

Measure the result after release

Keep the baseline, population and success measure attached to the change. Re-run the same lab journey, monitor field data as it matures and check the commercial measure the work was intended to support. Performance can regress gradually as content, apps and campaigns change, so make ownership and monitoring part of the recommendation.

If the audit needs to cover the wider platform as well as speed, see Webamondo’s ecommerce technical review or discuss ongoing Magento development.