Count the enquiries, not the visitors. Then write today's numbers down before you change anything, because a rebuild you cannot compare is a rebuild you cannot judge.
Most owners never do the second part. They redesign, the site looks better, and a year later nobody can say whether it made a dollar of difference.
What number actually matters?
Enquiries per month, and where they came from.
Traffic is an input. It is not a result. A site can double its visitors and bring in fewer jobs, which usually means the new visitors were the wrong people. A site can lose visitors and bring in more work, which usually means the remaining ones were the right people finding the right page.
So the headline number is the count of humans who called, emailed, filled in the form or booked. Everything else in your analytics exists to explain that number, not to replace it.
If your enquiries are flat while your traffic climbs, the problem is on the page and not in the search results. That specific failure has its own post: traffic but no leads.
Why do the numbers in analytics feel so slippery?
Because most of them measure attention, and attention is easy to misread.
Take engaged sessions, the metric that replaced bounce rate in Google Analytics 4. Google defines an engaged session as one that meets any of these: it lasts longer than 10 seconds, it has a key event, or it has 2 or more page or screen views (Google Analytics Help, "Sessions"). That is a low bar by design. Somebody who reads your opening paragraph for 11 seconds and leaves is counted as engaged.
This is not a flaw. It is just a reminder that the metric answers "did anything happen" and not "did this go well". Read it as a rough temperature, and never let it stand in for the enquiry count.
The same caution applies to average time on page and to session counts. They are useful for comparing this month with last month on the same site. They are close to meaningless as a target in themselves.
Which four numbers should you check each month?
Keep it to a page. Ten minutes at the start of the month is enough.
- Enquiries, counted by hand. Calls, emails, form submissions, bookings. One number, from your inbox and your phone, not from software.
- Where they came from. Search, Google Business Profile, social, a referral, someone who already knew you. Ask new customers directly. This is the only cheap way to learn which channel is actually paying for itself.
- Which pages people find you through. Google Search Console shows the queries and pages bringing you clicks. The report opens on the last three months by default (Google Search Console Help), which is a sensible window to skim.
- What happens on your one money page. The service page or product page that matters most. Are people reaching it, and are they going from it to your contact page?
Four numbers. If a metric is not on this list, you are allowed to ignore it this month.
Are you even counting the enquiries?
Ask this before you trust any dashboard, because the answer is often no.
Plenty of sites cannot report how many forms were submitted, for the simple reason that nobody ever set up an event to record it. The dashboard then shows a confident zero, or nothing at all, and the owner concludes the site is dead when it is merely uninstrumented.
Worse, some forms submit successfully and the message never lands in anyone's inbox. The visitor sees a thank-you screen, you see silence, and the site looks like a failure when it is a delivery problem. If your enquiry count dropped without an obvious cause, rule that out first: contact form emails that never arrive covers how.
The practical fix is unglamorous. For one month, keep a note on your phone. Every call, every email, every form. At the end of the month you will have a real number, and you will also know whether your software agrees with reality. If it does not, the software is what needs fixing.
Phone calls deserve their own line in that note. On a mobile screen a tap on your number is the most common way a local business gets contacted, and it is invisible in analytics unless someone deliberately tracked it.
What should you record before a rebuild?
Everything above, frozen, on the day before the new site goes live.
Export your Search Console performance data. Export or screenshot your monthly sessions. Write down your enquiries per month for the last twelve months, from your own records. Save the list of your top pages and top queries. Put it all in one folder with the date on it.
This takes an hour and it is the difference between an opinion and an answer. Without it, the only available verdict on a redesign is "it feels better", which is exactly the verdict everyone gives about a site they just paid for.
The same folder protects you in the other direction. If traffic falls after launch, the before-picture is what lets you tell a redirect problem from normal seasonal noise, and the recovery steps are in redesigning without losing rankings.
How long before you judge the result?
Two weeks is long enough to catch breakage. It is far too short to judge performance.
Split it in your head. In the first fortnight you are checking that nothing is broken: forms arriving, email flowing, pages loading, old URLs redirecting. That is a maintenance check and it needs doing immediately.
Judging whether the site works is a different exercise and it needs a season. Search results take time to settle after any structural change. Your own trade has a rhythm, and a roofer comparing November with June learns nothing about their website. Compare a quarter against the same quarter last year wherever you can, and be honest about anything else that changed in the meantime, like an ad campaign starting or a competitor closing.
One thing to check before any of this
Whose account is the data in?
If your analytics and Search Console sit inside an agency's login or a former developer's personal Google account, none of the above is really yours, and a baseline you can lose is not a baseline. Confirm you are an owner on both properties, in your own account, today. The full list of logins worth auditing is in who actually owns your website.
Where to start this week
Open a note and write three lines: how many enquiries you got last month, where they came from, and which page on your site you most want people to reach. If any of the three makes you pause, that is your answer about what to fix first, and it costs nothing to find out.
For sites we build, this measurement is part of the work rather than an add-on, and what sits inside each scope is set out on web design, with the ranges on pricing. Keeping it watched month to month is what the care plan is for, from $129 per month.
If you want an outside read on whether your current site is earning its keep, send us the URL for a free teardown. We will tell you what it does well, what is quietly leaking enquiries, and which single number we would watch if it were ours, whether or not you work with us.