Start with the business decision and a clean baseline
A conversion audit needs a defined outcome. For a service business, the main action may be a qualified quotation request, booked consultation, phone call or WhatsApp conversation. For an ecommerce site, it may be a completed purchase. Choose one primary action for the page being reviewed and treat supporting actions—such as brochure downloads or video views—as journey evidence rather than equal successes.
Record the current position before changing copy or layout. Use a representative period and note visits to the relevant page, completed primary actions, qualified leads, sales outcomes and obvious sources of noise such as spam. Calculate the page conversion rate as completed primary actions divided by relevant visits, but keep the raw counts and lead quality beside the percentage. Ten excellent enquiries and ten unsuitable enquiries are not the same result.
Do not begin with an industry benchmark. Traffic source, offer value, brand familiarity and sales cycle can make two apparently similar websites behave very differently. The useful comparison is the site's own stable baseline, segmented by page, device and meaningful acquisition channel.
- Primary action and what makes it commercially useful
- Visits, completed actions and qualified outcomes for the same period
- Mobile and desktop results reviewed separately
- Known tracking gaps, promotions or seasonal effects
Test whether the page makes sense before it tries to persuade
Open the page as if you know nothing about the company. Within the first screen, a suitable customer should be able to identify what is offered, who it is for, the useful outcome and the next step. A clever headline can support the brand, but it should not force visitors to decode the service before they can decide whether to continue.
Read each section in order and note every unanswered buying question. Common gaps include service scope, who the work suits, location or delivery model, timing, price factors, process, proof and what happens after an enquiry. Add information where it reduces a real decision barrier; do not add paragraphs merely to make the page longer.
Check that calls to action describe the result of clicking. 'Request a website assessment' is clearer than 'Submit', and 'See relevant case studies' is clearer than 'Learn more'. Keep the primary action visually consistent, while providing a lower-commitment route such as a guide or case study for visitors who still need evidence.
Review trust as evidence, not decoration
Trust grows when the page answers 'Who is behind this?', 'Can they do the work?' and 'What will happen to my enquiry?' Use accurate company details, real people, an understandable process, relevant work examples and authentic public feedback. Place proof close to the claim it supports instead of isolating every testimonial in a distant carousel.
Match the proof to the decision. A technical service benefits from a clear explanation of the approach and measurable case-study context. A local service benefits from accurate service-area information and genuine customer feedback. An unfamiliar agency benefits from visible ownership, contact details and a straightforward commercial process.
Remove anything that creates doubt: unsupported badges, anonymous claims without context, inconsistent phone numbers, old dates presented as current, broken social links or results without a stated period and measurement basis. Honest limitations make credible evidence stronger; they do not weaken it.
- Real company and contact information
- Proof relevant to the service on the page
- Clear process, responsibilities and next step
- No invented scarcity, guarantees or unsupported partner claims
Walk the complete journey on mobile and desktop
Use a real phone, not only a narrowed desktop window. Start from the page a search advert, organic result, social post or referral would open. Check the headline, navigation, proof, call to action, form, success message and return path. Repeat the journey on desktop because layouts, menus and contact behaviour can differ.
Look for interaction costs: a menu that covers its final links, a floating button that hides a form, text that requires zooming, horizontal scrolling, unstable content or a call to action that disappears below an oversized hero. Verify that phone, email and WhatsApp links open the intended destination without forcing the visitor to copy information manually.
Test slower connections and ordinary devices where possible. Google describes Core Web Vitals as real-world measures of loading performance, responsiveness and visual stability. Lab tools help diagnose a page before release, while field data better represents actual visitors; neither score should replace testing the business journey itself.
Audit forms and contact routes for preventable friction
Ask only for information needed to respond or qualify the next step. Every extra required field creates work and should earn its place. Explain why sensitive or unusual information is requested, use the correct input type and autocomplete value, and show an example when the expected format is not obvious.
Every form control needs a visible, meaningful label. The W3C Web Accessibility Initiative recommends labels that identify text fields, checkboxes, radio buttons and select controls. Instructions, validation and success or error feedback should be understandable without relying only on colour.
Submit a controlled test only when the business has approved it. Confirm that one successful action creates one message, reaches the correct inbox or lead system, shows a useful confirmation and does not expose submitted details in the URL. Check spam handling and response ownership as well as the front-end form; a fast form cannot rescue an enquiry that nobody receives.
- Required fields are necessary and clearly labelled
- Errors explain what needs to be corrected
- The submit state prevents accidental duplicate requests
- Success is confirmed to the visitor and the responsible team
Separate usability evidence from measurement evidence
Analytics can show where people arrive, continue and complete a tracked action, but it cannot explain every hesitation. Combine quantitative evidence with direct journey testing, sales feedback, customer questions and—where the volume justifies it—privacy-conscious usability research.
Confirm that the primary action is measured once and only after success. Keep secondary interactions available for diagnosis, but do not let scrolls, button clicks or partial form starts inflate the headline conversion total. Reconcile website events with delivered enquiries and qualified opportunities so reporting reflects business reality.
Treat missing data as a limitation, not permission to guess. Small sites may not have enough volume for a statistically confident split test. In that case, fix clear defects first, document the change and compare a longer period while accounting for seasonality, traffic mix and campaign changes.
Turn findings into a thirty-day improvement plan
Score each issue by customer impact, supporting evidence, confidence and effort. Repair broken forms, unreadable mobile layouts, hidden calls to action and tracking duplication before debating cosmetic refinements. Group related changes so the team can explain what was altered and why.
Make the smallest change that can resolve the identified problem. If customers cannot tell who the service is for, improve the proposition before commissioning a full redesign. If qualified visitors abandon a long form, remove unnecessary fields before buying more traffic. If leads are arriving but not converting, investigate the offer and follow-up process instead of assuming the page is solely responsible.
After release, confirm that the journey still works and annotate the date. Review primary actions, qualified leads, device patterns and customer feedback against the baseline. Keep improvements that help the business outcome, investigate ambiguous movements and reverse changes that introduce new friction.
- Week 1: fix broken journeys, contact details and measurement defects
- Week 2: strengthen message clarity, proof and action labels
- Week 3: improve mobile interaction, forms and performance bottlenecks
- Week 4: verify delivery, compare evidence and choose the next test