Website analytics: the numbers that actually matter
Analytics can bury you in numbers that never change a decision. For most small businesses, a handful of measurements tell you almost everything you need to know, and the rest is noise. This guide covers which ones to watch, how to set up Google Analytics 4 simply, how to tag your links so you know what marketing is working, and how to do all of it while respecting visitors' privacy.
Decide what success is first
Analytics only answers questions you have asked. Before opening any report, write down what a good visit looks like for your business: a call, a form, a booking, a purchase. Those are your conversions, and everything else is context for them.
The numbers worth watching
- Conversions: the calls, form submissions, bookings, or sales that reflect real business value. This is the real goal.
- Traffic sources: where visitors come from, such as search, social, email, referrals, or direct.
- Top pages: which pages actually pull people in, and which ones lead to enquiries.
- Device split: how mobile and desktop visitors behave differently, because they often convert differently.
- Core Web Vitals: whether the site is fast enough, which Search Console reports.
The numbers to ignore, mostly
Raw pageviews and bounce rate in isolation rarely change what you do. A page can have a high bounce rate because it answered the question in one visit, and a high pageview count can mean people are lost. Focus on whether visitors turn into customers and where the good ones come from.
Setting up Google Analytics 4
Google Analytics 4, or GA4, is the free tool for understanding how people use your site. It can feel overwhelming, but you need only a few things from it.
- Install it on every page of the site.
- Tell it which actions count as key events (called conversions in earlier versions): form submissions, call and email taps, booking completions, and purchases.
- Review a short list of reports regularly: where visitors come from, what pages they view most, and how many key events happened.
You do not need every report. You need the handful that inform decisions, checked on a schedule.
Conversion tracking: measure the actions that matter
Without conversion tracking you cannot tell what is working or where to improve.
- Form submissions and enquiries.
- Phone calls and click-to-call taps on mobile.
- Bookings or purchases.
- Key steps on the way to a sale, such as starting a checkout or opening a quote form.
Keep it simple. A short list of conversions that genuinely reflect business value, reviewed regularly, beats a complex setup that nobody looks at.
UTM tags: know which marketing worked
UTM parameters are small tags added to the end of a link so your analytics knows exactly what sent a visitor: which email, ad, or post. They turn "traffic" into "traffic from this particular thing".
- utm_source says where it came from, such as a newsletter or a social network.
- utm_medium says the type, such as email, social, or paid.
- utm_campaign names the specific campaign or post.
- Use consistent, lowercase names, or the same source will be split across several entries.
- Use them on links you place elsewhere, such as emails, social posts, and ads. Do not put them on links between pages of your own site, because that can break the attribution of the original visit.
Without tagging, you are guessing which marketing works. With it, you can do more of what produces enquiries and stop spending effort on what does not.
Measure with consent
Use privacy-respecting analytics and honour cookie consent, which in many places is a legal requirement rather than a courtesy. You can learn what you need without compromising visitors' trust. Be aware that where visitors decline consent, your figures will undercount, so treat them as a reliable trend rather than an exact tally.
A routine that works
- Weekly, two minutes: glance at conversions and traffic sources for anything unusual.
- Monthly, fifteen minutes: top pages, which sources produced conversions, and what changed since last month.
- Quarterly: decide one thing to change based on what the numbers said, and check afterwards whether it worked.