Setting up analytics that actually tell you something

Analytics installed straight out of the box tells you how many people visited and roughly where they came from. Neither answers a business question.

To get value you need to configure it around what actually matters to you. Here's the minimum.

Configure: conversions

The single most important step, and the one most often skipped.

You need to tell the tool what counts as success. For most service businesses that's contact form submissions, phone number taps, and email address clicks.

Without this, you can see that 400 people visited and nothing about whether any of it mattered. With it, you can see that the ten visitors from one search term produced three enquiries while the two hundred from social produced none.

That single comparison changes budgets.

Configure: phone tracking

If a meaningful share of your enquiries come by phone — true for most local service businesses — and you're not tracking calls, you're missing most of your results.

At minimum, track taps on the tel: link. That undercounts (people also copy the number, or call from Google directly) but it's far better than zero.

For real accuracy, call tracking numbers show which channel produced each call. Worth it if you spend meaningfully on ads.

Configure: filter yourself out

You and your team visit your site constantly. On a low-traffic site that's a serious distortion.

Filter internal traffic by IP address. Otherwise you'll see a spike every time someone in the office checks a page.

Configure: link Search Console

Free, takes two minutes, and adds the thing analytics can't tell you on its own: what people searched to find you.

This is often the most useful data you'll have. You'll find terms you never thought to target and pages ranking for things you didn't intend.

Read: the four reports that matter

Traffic by source. Which channels bring people, and — with conversions configured — which bring people who act. These two lists are frequently very different, and the gap is where budget decisions live.

Landing pages. Which pages people arrive on. Usually a small number carry most of the entry traffic, and those deserve disproportionate attention.

Conversions by source. The only report that answers "is this working." Look at conversions, not sessions.

Search Console queries. What people typed. Sort by impressions with low click-through: those are terms where you appear but nobody clicks, usually meaning your title and description need work. That's often the cheapest win available.

What to ignore

Bounce rate, mostly. GA4 changed how it works and it's widely misread. Someone reading your entire article and leaving satisfied isn't a failure.

Time on page. Poorly measured and easily misinterpreted.

Total pageviews as a goal. Traffic is not the objective. Enquiries are.

The honest limitation

Attribution is imperfect and getting more so. Privacy features, blockers, and cross-device journeys all break the chain. Someone might find you on their phone, think about it for a week, then search your name on a laptop and call.

Analytics will credit that to "direct" and tell you nothing about the search that started it.

So use it directionally. It's good at telling you which channels are broadly working and which pages people enter through. It's bad at precisely apportioning credit. Treat confident attribution claims with suspicion — including from anyone selling you marketing.