Reading Your Website Analytics Without Fooling Yourself

Every website has analytics, and almost every business looks at them monthly, notes whether the line went up, and closes the tab. The ritual is harmless but it is not analysis, and the gap between the two is where a lot of confident, wrong decisions get made.

The problem is not that the numbers are inaccurate. It is that they answer a narrower question than the one being asked of them.

What the numbers actually mean

Visits are not people. One person on a laptop, a phone, and a work machine is three visitors. Someone who clears their browser data is a new visitor every time. A month-on-month rise in "users" can be entirely composed of the same people arriving differently.

Time on page is mostly fiction. It is usually calculated as the gap between loading one page and loading the next, which means the last page of any visit — very often the most important one — contributes nothing. A page that people read carefully and then leave from looks identical to one they abandoned instantly.

Bounce rate punishes success. If someone arrives, finds your phone number, and calls you, that is recorded as a bounce. The single-page visit that achieved exactly what the page existed for is filed with the visits that achieved nothing.

None of this makes the tools useless. It makes them instruments with known distortions, which is fine as long as you know the distortions.

The bias that matters most

Your analytics systematically over-count the people who least resemble your customers.

Bots, scrapers, uptime monitors, link previewers, and your own team account for a substantial share of traffic on a small site. Internal traffic is the worst offender because it is high-engagement by nature: your own people visit repeatedly, load many pages, and stay a long time. Left unfiltered, the profile of your "most engaged visitors" is a portrait of your staff.

Excluding internal and automated traffic is the single highest-value thing most small businesses can do to their analytics, and it usually takes an afternoon. Until it is done, every other conclusion is drawn from a sample that includes the people who already know everything.

Speed is a measurement, not a vanity metric

One category of number does deserve direct attention, because it affects everything else you measure.

The Web Vitals initiative exists "to provide unified guidance for quality signals that are essential to delivering a great user experience on the web" [1], and the reason it matters analytically is subtle: a slow page loses visitors before they are counted as having done anything. Those people are recorded as arrivals with no engagement, which looks like a content problem and is a performance problem.

This is one of the few places where a technical measurement maps cleanly onto a business outcome, and where improving the number reliably improves the thing.

Collect less, on purpose

The instinct when analytics disappoint is to add more tracking: more events, more custom dimensions, another tool alongside the first.

This reliably makes things worse. Each addition increases what you must maintain, slows the page you were trying to improve, and produces more numbers nobody has time to interpret. The rule worth applying is the one we hold ourselves to for data generally — not collecting it speculatively "in case it is useful later" [2]. An event you have never used to make a decision is not insight; it is cost with a chart attached.

There is also a legal dimension that is easy to drift past. Analytics data about identifiable visitors is personal data, and the people it describes have the right to receive it "in a structured, commonly used and machine-readable format" and to transmit it elsewhere [3]. That right is much easier to honour when you have collected a small, deliberate set of things than when you have accumulated whatever three tools happened to offer.

The four questions worth answering

Most businesses need far less than they collect. Almost all of the value is in four questions.

How do people arrive? Search, direct, referral, social, email. Not for its own sake, but because it tells you which of your efforts produce visits at all.

What do they look at first? Your homepage is often not the entrance. If most visitors land on one specific page, that page is your front door, whatever the navigation says.

Where do they stop? The page after which the visit ends, and — much more usefully — the step of a form where people abandon. Form abandonment is the most under-examined data most sites hold, and the most immediately actionable.

How many did the thing? Booked, enquired, subscribed, purchased. Everything else is a proxy for this.

If a number does not help answer one of those four, it is decoration.

Averages hide the thing you are looking for

One habit does more damage than any particular metric, and that is reading averages.

An average session duration of ninety seconds might describe a site where everyone stays ninety seconds. It might equally describe one where nine people leave immediately and one person reads for fifteen minutes. Those are entirely different websites with an identical number, and the second one contains a finding the first does not.

The same applies to conversion rates across a whole site, average pages per visit, and any figure aggregated over traffic sources that behave nothing alike. Visitors from a search for your company name and visitors from a paid advertisement have almost nothing in common, and averaging them produces a number describing nobody.

The correction is to split before you summarise. Look at each significant source separately, and look at the shape of the distribution rather than its middle. Most of the useful findings in small-business analytics are in a segment that represents a fifth of the traffic and behaves completely differently from the rest — which is invisible in any average and obvious the moment you separate it out.

A better monthly ritual

Instead of asking whether traffic went up, pick one question and answer it properly.

Traffic rising while enquiries stay flat is not a success; it is usually a sign that new arrivals are less relevant than the old ones. Traffic falling while enquiries hold is often fine. The relationship between the two is the story, and neither number tells it alone.

Then ask what you would do differently if the number were 20% higher or lower. If the answer is nothing, stop reporting it. That single test removes most of what fills a typical monthly report, and what remains is worth the meeting.