How to track website traffic

Most guides to tracking website traffic start by listing metrics. That is the wrong end. The useful starting point is that traffic data only earns its keep when it changes something you do, and almost all of it never will.

So this page works backwards: which numbers actually change decisions, what you need installed to see them, and the setup mistakes that make the whole exercise worthless later.

The five numbers that change decisions

Unique visitors tells you whether the site is growing. Watch it monthly, not daily; daily movement is mostly noise.

Top pages tells you what people actually want, which is reliably different from what you assumed when you planned the site. This is the single most useful report for anybody writing content, and it is the one people check least.

Traffic sources tells you which of your efforts are working. Split into search, direct, referral and social at minimum, and watch for referrers you did not expect: a link from somewhere you have never heard of is often the most actionable thing in the dashboard.

Time on page or engagement tells you whether people who arrive are staying. A page with high traffic and low engagement is usually a headline writing a cheque the content does not honour.

Conversions tells you whether any of it matters. Define one goal, even a soft one like reaching the pricing page, before you look at anything else. Traffic without a goal is a vanity number.

Getting measurement running today

For a modern tool this is a genuinely small job. You paste one script tag into the head of your pages, or install a plugin if you are on WordPress, Shopify, Squarespace or similar, and data starts arriving immediately.

The lightest tools are around one to three kilobytes and need no build step, no npm package and no tag manager. Google Analytics 4 is the outlier: it is tens of kilobytes, usually goes in through Google Tag Manager, and needs configuration of events, conversions, consent mode and data streams before it tells you much. That configuration burden is why a consultancy industry exists around it.

If you are in the UK or EU there is one more decision before you paste anything, and it is the one that matters most.

Decide the cookie question before you install

A tool that sets cookies needs a consent banner, and every visitor who declines is simply missing from your data. They are not estimated or modelled. They are absent.

This is the largest accuracy problem in most analytics setups and it is invisible from inside the dashboard. A cookieless tool needs no prompt, so it counts everybody, and it is blocked by ad blockers far less often than a recognised advertising tag.

Decide this first, because changing it later means a second migration and a break in your history.

The four mistakes that ruin it later

Not filtering your own visits. Your team reloading the site all day will quietly distort every engagement metric you have. Most tools handle this; almost nobody switches it on.

Not counting bots. Around a third of raw arrivals at a typical site are automated: declared crawlers, headless browsers, scrapers from datacentre ranges, link previews nobody saw. Tools differ enormously here, and some meter you for all of it. Check whether your tool filters bots before counting, and before billing.

Never defining a goal. Six months of traffic charts with nothing to compare them against is six months of decoration. Define one conversion on day one.

Changing tools without overlapping. Always run the new tool alongside the old one for a month. The two will not agree, and understanding why is more valuable than either number alone.

The rest of the field

Common questions

Related guides