A project page may draw hundreds of visits and still produce no useful enquiries. Before calling it a failure, check whether visitors can quickly find the building type, location, scope of work and a way to contact the practice. Traffic alone cannot tell you whether the wrong people arrived or qualified clients left with unanswered questions.
Start with decisions, not a dashboard
Website performance means something different to an architecture practice, a property developer and a public agency. First, write down what the site needs to help people do. A practice might want potential clients to find its healthcare portfolio; a developer might need visitors to find available floor plans. A public project team might need residents to locate consultation dates and accessible documents.
For each priority, choose an outcome and a few measures that help explain it. A click is not a useful target simply because it can be counted.
| Decision to support | Useful measure | Important qualification |
|---|---|---|
| Is a service page reaching relevant visitors? | Visits by source, search query and location | High volume may come from students or job seekers rather than clients |
| Can prospects make an enquiry? | Completed forms or tracked contact actions | A click on an email address does not prove a message was sent |
| Can the public find a project document? | Document downloads and navigation paths | A download does not prove the document was read or understood |
| Does a page work on mobile? | Mobile task completion, errors and load experience | Device averages can conceal problems on a specific page |
These measures are clues, not guarantees of commissions, leases or public understanding. A large civic project may attract considerable interest without having a commercial enquiry route.
Make the data dependable
Define events precisely
Decide what counts as an enquiry before setting up analytics. Record a completed form only after a successful submission—not when someone opens it or presses a button that returns an error. Name other meaningful actions separately, such as a project brief download or consultation registration. For each event, document its definition, location on the site and test procedure. That way, a later website change is less likely to alter the figures unnoticed.
Test the full process on desktop and mobile. Send a test enquiry, check that it reaches the intended inbox or system, and confirm that the analytics event fires once. Keep names, email addresses, message text and confidential project details out of event fields. Check that privacy notices and consent settings match the tools in use; where consent limits measurement, label the figures accordingly.
Separate different kinds of evidence
Analytics shows recorded visits and actions, not every visitor’s intent. Search performance data may show which queries brought a page into view. Enquiry records can tell you whether a lead concerned a real project; a short usability test may reveal why people missed a floor-plan link. Keep those findings distinct. An analytics conversion rate is not a measure of signed work.
On property-related pages, distinguish market context from your own site’s results. Public resources such as HUD's housing program information may help explain a regulatory or housing-policy context, but they cannot validate your traffic or the quality of your enquiries.
Read patterns at the right scale
Look beyond the homepage to the pages that serve important tasks. Group them by purpose—services, built projects, active developments, public consultations and practical resources—before making comparisons. A completed residential project and a time-limited consultation page differ in audience, lifespan and purpose. For a small practice, a handful of genuine enquiries may matter more than a percentage change in visits.
Use consistent date ranges and note events that could affect the numbers: a project publication, planning announcement, event, redesign or temporary campaign. Compare periods only when circumstances are reasonably similar. A rise in visits after local news coverage may be significant, but it does not prove that a page edit caused it.
Averages can hide a specific problem. Compare mobile with desktop, relevant service areas with distant traffic, and key landing pages with the site overall. If the contact form performs poorly only on phones, test the form, keyboard behaviour and confirmation message before rewriting the service page. Treat small counts carefully: two enquiries becoming four warrants a closer look, not a claim of lasting doubled performance.
Turn a report into a controlled improvement
A monthly review can fit on one page: the question, the evidence, the proposed change and when to check the result. Rank issues by their effect on an important task and the work needed to fix them. A broken document link during a live public consultation needs attention sooner than a modest shift in average time on an older portfolio page.
- Identify a specific gap. For example, a healthcare service page gets relevant visits, but few visitors reach its project examples.
- Check the page directly. Inspect navigation, link labels and mobile layout. Ask someone unfamiliar with the site to find a relevant project.
- Change one clear element. Add a plainly labelled link to healthcare projects rather than replacing the page, form and navigation at once.
- Verify and review. Check the link, record the change date and compare subsequent behaviour with a suitable earlier period.
If a website helps explain a building before occupation, assess whether people can find essential information, not just whether they visit. An article on urban noise control in building design, for example, belongs on a project page only if it helps visitors understand a relevant design decision.
Keep the reporting proportionate
Give someone responsibility for the measurement plan and revisit it when services, projects or site features change. Keep a short log of event definitions, known data gaps and major edits. Restrict access to analytics and enquiry records to those who need it, and set retention practices in line with applicable requirements and organisational policy.
At the next monthly review, pick one priority page and follow its full task path on a phone: arrival, key information, contact or document action, and confirmation. Note the first point that becomes unclear. Fix and retest it before adding another dashboard metric.