Internet traffic tracker: the four kinds, what each one can count, and how to choose

An internet traffic tracker records who visits a website and what they do, and there are four ways to do it: a JavaScript tag in the page, a server side collector, a reader of raw server or CDN logs, and a panel based estimator that models traffic to sites you do not own. Tag based tracking is what most people mean and what most people install. It is also the one with the largest blind spots, which is why two trackers on the same site report different numbers and both are telling the truth about what they can see.

Pick by the question you need answered. For your own site and your own marketing decisions, a tag plus your server logs covers almost everything. For somebody else's site, only the panel based estimators are available to you, and their output is a model rather than a count.

The four kinds, side by side

Kind	How it works	Sees	Misses
Tag based	JavaScript in the page fires on load	Page views, events, referrer, device, journey through the site	Blocked scripts, visitors who decline consent, non JavaScript clients, most bots
Server side	Your server sends the event to the collector	Everything the server handles, including blocked browser cases	Behaviour that never reaches the server, such as scrolling
Log based	Parses raw access logs from the server or CDN	Every request, including bots, crawlers and files	Who the person is, and anything served from cache beyond the log
Panel based	Models from clickstream panels and public signals	Rough traffic for any domain, including competitors	Precision, and small sites entirely

What a tag based tracker cannot see, and how large the gap is

Three groups of visitors are routinely missing from tag based numbers. Visitors with an ad or tracker blocker, who never run the script. Visitors who decline consent in a region where tracking requires it, which is covered by the PECR cookie rules in the UK and the equivalent rules elsewhere; a plain language summary of the consent requirement sits in this guide to cookies under GDPR. And clients that do not execute JavaScript at all, which includes most automated agents.

The gap varies enormously by audience: a site selling to developers loses a far larger share than a site selling garden furniture. The way to size your own gap is to compare the tag count against your server logs for the same week. The ratio is stable enough to use as a correction factor, and knowing it is worth more than switching tools.

Choosing a tracker for your own site

• Google Analytics is the default, free at the volumes most sites run at, and integrates with Search Console and Google Ads. Setting it up takes under an hour. Its own documentation on metric differences is the honest explanation of why a session count here never matches another tool exactly.
• Privacy first tools such as Plausible and Matomo are lighter, simpler and in many configurations avoid the consent banner entirely. Matomo can be self hosted, which keeps the data on your own infrastructure.
• Edge analytics such as Cloudflare Web Analytics count at the network layer, which sidesteps a chunk of the blocking problem if your traffic already passes through that network.
• Log analysers such as GoAccess read the raw access log. This is the only view that includes crawler and bot traffic, which makes it the right tool for technical questions rather than marketing ones.

A reasonable setup for most businesses is one tag based tracker for marketing questions and a periodic log review for technical ones. Running two tag based trackers in parallel produces two sets of numbers and a recurring argument, so if you do it, decide in advance which one is the source of truth for reporting.

Three practical decisions, and the tracker you need for each

1. Which page earns the most organic conversions? Tag based tracker, landing page report, filtered to organic search, with a conversion defined. Nothing else answers it. The method is covered in measuring organic traffic properly.
2. Is a crawler hammering the site? Log analyser. User agent and request frequency over time, which no tag based tool can give you because bots do not run JavaScript.
3. Is that competitor bigger than us? Panel based estimator, read as a range rather than a number. Reading those estimates properly matters more than which vendor you use, and the source of the numbers is worth understanding before you quote one in a board pack.

Setting one up so the data is usable in six months

Three decisions made on day one save a lot of work later. Define your conversions before you need them, because a conversion defined in March has no data for January. Agree a naming convention for campaign tags and write it down, since inconsistent UTM values are the single most common cause of unreadable channel reports. And exclude your own team's traffic by IP or by a browser flag, so the internal review session does not land in the same bucket as a customer.

Then set a retention policy that matches how you work. Most reporting questions reach back thirteen months, for a year on year comparison, and anything older is for trend charts rather than decisions. Export a monthly summary to a spreadsheet you control, so a tool change or a plan downgrade never costs you the history.

Counting, cookies and consent: the tracker questions

Why do two traffic trackers report different numbers for the same site?

They count different events with different rules: what starts a session, how a bounce is defined, whether bots are filtered, and whether the visitor consented. Pick one as the source of truth for reporting and use the other for the question it answers best.

Is a traffic tracker allowed without a cookie banner?

It depends on what the tracker stores and reads on the device. Tools designed to avoid storing identifiers on the visitor's device are commonly run without a consent banner, and the regulator guidance linked above sets out the test. Check the specific configuration you are running with whoever handles your privacy compliance.

How much traffic do blockers hide?

Enough to matter, and the share depends entirely on your audience. Measure your own gap by comparing a week of tag data against a week of server logs, then apply that ratio when you talk about absolute numbers.

Can I track visitors from a specific company?

Reverse IP lookup services match some visits to company names, and they work better for large organisations on corporate networks than for remote workers on home connections. Treat the output as a lead signal rather than a visitor count.

What is the best free website traffic tracker?

Google Analytics for breadth and integration, a privacy first tool for a lighter setup with a simpler report, and a log analyser for the technical view. All three have free paths, and most sites genuinely benefit from running two of them.

How long should traffic data be kept?

Keep at least thirteen months live so year on year comparisons work, and export a monthly summary into a spreadsheet of your own for anything older. The export protects you from retention settings and plan changes.

Visits your tracker never records, from people who already decided

There is a category of attention that reaches none of the four kinds above. Somebody asks ChatGPT, Gemini, Perplexity or Claude for a recommendation in your category. The assistant reads a few pages, writes a short answer, names two or three companies and the person acts on it. If you were named and they went straight to your pricing page, the tracker records a direct visit with no referrer and no campaign tag. If you were not named, the tracker records nothing at all, and nothing in a traffic report distinguishes the two.

That makes it a measurement gap rather than a marketing theory: a decision was influenced and the log shows none of it. The only way to see that surface is to ask the assistants the questions your buyers ask and read what comes back. AI Knows Us counts what the assistants say about you, which is the one number no tracker on your site can report, and it keeps the answers over time so you can see the direction. The direct traffic line in your report usually makes more sense afterwards.