A website can feel faster while generating the same number of useful enquiries. It can also receive more enquiries because a campaign changed, even when its technical performance stayed the same. Both outcomes matter, but they answer different questions. A useful measurement plan keeps the experience, the customer action and the business result distinct.
Understand the three experience signals
Google's current Core Web Vitals guidance covers loading, responsiveness and visual stability. Its good thresholds are Largest Contentful Paint at 2.5 seconds or less, Interaction to Next Paint at 200 milliseconds or less, and Cumulative Layout Shift at 0.1 or less, assessed at the 75th percentile and separated by device type. The official Web Vitals reference explains these measures. Recheck it when planning a future audit because metrics can evolve.
Use a repeatable technical test to identify issues such as oversized photographs, delayed content or a button moving as the page loads. Keep the device profile, network setting and tested page consistent. A single test is an observation under those conditions; it is not a complete account of every customer's experience. Pair laboratory tests with field data from actual visits where available. Lighthouse does not measure INP in a standard page-load test without interaction; its Total Blocking Time measure is a diagnostic proxy, not an INP result.
Define the action you actually care about
Write a small measurement dictionary. A service-page visit, an enquiry-button click, a submitted form and a qualified lead are different events. If a WhatsApp button opens a conversation, the website may know the click occurred but not whether the visitor sent a message. Label the report honestly instead of counting every click as an enquiry.
Agree what makes an enquiry relevant before examining results. For example, a project brief might need a suitable service, a reachable contact and a realistic need. Keep that definition stable during the review. Do not improve the apparent conversion rate by quietly excluding inconvenient visits or redefining a qualified lead halfway through.
A worked example with clear denominators
Suppose an illustrative service page receives 1,000 visits in a month. Forty of those visits include a contact-action click, twenty produce an identifiable enquiry and twelve produce an enquiry meeting your agreed relevance criteria. Count each type of outcome at most once per visit, and include only outcomes matched to those visits in this example. The contact-action rate is 40 divided by 1,000, or 4 percent. The observed enquiry rate is 2 percent. The relevant-enquiry rate is 1.2 percent. These numbers are examples for explaining the calculation, not results from YAMAR or a client.
Next month, using the same definitions and counting method, there are 1,200 visits and eighteen visits that produce relevant enquiries. The relevant-enquiry rate is 1.5 percent. That change alone does not prove a speed improvement caused it. Check whether traffic sources, offers, seasonality or sales handling changed. If the sample is small, describe the result as an early observation and continue collecting comparable data.
Prioritise work around visible friction
Review the actual customer route while reading the numbers. A heavy hero video might delay the service explanation. A form might be technically fast but ask for information a visitor does not have. A successful submission might reach an unattended inbox. The appropriate repair depends on where the problem occurs, so connect each proposed change to an observed issue.
Keep a change log with the release date, affected pages and reason for the work. Improve a small, coherent set of problems, then review the same measures. If several campaigns and design changes launch together, acknowledge that their individual effects are difficult to separate. Good reporting makes uncertainty visible.
A weekly review checklist
- Check key pages on a phone and a desktop.
- Repeat performance tests under documented conditions.
- Compare relevant periods and identify traffic-source changes.
- Distinguish clicks, submitted enquiries and qualified leads.
- Investigate failed forms and missing handovers.
- Record changes to offers, campaigns and page content.
- Assign one owner and one next action to each issue.
- Keep personal customer information out of public reports.
Report what the evidence supports
A useful report might say: the service image is smaller, the test's loading measure improved, and the enquiry rate is being monitored. That is more credible than declaring a revenue increase before sales records exist. Avoid turning a technical score into a business guarantee. Performance work can remove friction, but pricing, relevance, reputation and response quality still affect customer decisions.
YAMAR plans website foundations and enquiry journeys together. For an improvement project, bring the pages you want to assess, your current reporting and the way your team handles enquiries. We can define the measurement gaps and scope a first set of changes. The proposal should identify what will be tested, which results can be observed and who owns review after launch.
Source review: 4 November 2026. The worked figures are illustrative planning examples, not YAMAR or client results.