What analytics can and cannot tell you
Website analytics are reports about how people use a site. A tool records activity such as page requests, then summarizes it into charts and tables. Used well, those summaries help you notice what visitors read, where they come from and where they get stuck.
They are not a precise count of real people. Some visitors block measurement, some tools count differently from others, and automated programs can inflate figures. Think of analytics as a rough instrument that is good for spotting direction and large differences, and poor at settling small ones.
The basic terms
Names differ between tools, so check the definitions in whichever one you use. These are the common ideas.
- Visits or sessions: a group of actions by one browser within a period of activity. If someone returns the next day, that is normally a new visit.
- Visitors or users: an estimate of distinct browsers or devices. One person using a phone and a laptop may appear as two. Two people sharing a device may appear as one.
- Page views: the number of times pages were loaded. A visitor who reads three pages creates three page views.
- Pages per visit: page views divided by visits. A higher number is not automatically better. On a reference page, a quick single-page visit may mean the person found the answer immediately.
- Time on page or visit length: often estimated. Many tools cannot measure the time spent on the last page of a visit, so treat it carefully.
- Bounce or engagement measures: ideas about whether visitors did anything beyond landing on a page. Definitions vary between tools and over time.
Where visitors come from
A sources report groups visits by how people arrived. Typical groups include:
- Search: people who clicked a result in a search engine.
- Direct: people who typed the address, used a bookmark, or whose source could not be identified. Direct is often a catch-all.
- Referral: people who followed a link from another website.
- Social: people who came from social platforms.
- Email: people who clicked a link in a message, if links are labeled so the tool can tell.
- Campaign labels: links you tag yourself, which let you tell apart, for example, two different newsletters.
Do not treat the labels as perfect. Some visits are misfiled, and private browsing or apps can hide the source. If you send a newsletter, tagging links consistently makes the email part of your report more reliable. Understanding where people arrive from also connects to search intent: someone who finds a page by searching a specific question expects an answer to that question.
Which pages are viewed
The pages report shows what visitors read most. It is often the most practical report for a small site. Use it to ask:
- Which pages bring people in? These are your landing pages, so make sure they are accurate and have clear next steps.
- Which pages are rarely visited? They may be hard to find, poorly named or simply not wanted.
- Does anything surprise you? A page you barely promoted may be doing well, which tells you something about what readers value.
Pages that attract steady attention are good candidates for careful upkeep. Keep them current and free of broken links.
Do not over-read small numbers
This is the most important habit. With a small site, daily figures are tiny and swing a lot by chance. Suppose, as a hypothetical example, a page had 8 visits one week and 12 the next. That looks like a 50 percent increase, since 12 minus 8 is 4, and 4 divided by 8 is 0.5. But a change of four visits could easily be a single person revisiting, a few friends, or an automated program. It says almost nothing about a trend.
Some ways to stay level-headed:
- Compare longer periods. Look at months rather than days, and compare with the same period in a previous year if you can, because many topics are seasonal.
- Look for patterns over several periods. One high week is a data point. Several in a row are a pattern.
- Beware of percentages on small bases. A jump from 2 to 4 is a 100 percent increase, but it is still only two more visits.
- Remember your own visits. Your own browsing, and that of your team, can distort the picture unless you exclude it.
- Allow for bots. Odd spikes at strange hours or from unexpected places may be automated traffic.
- Do not chase single numbers. Optimizing for a metric alone can make the site worse for people.
Turning numbers into decisions
Analytics are most useful when you start with a question. A few examples:
- Are visitors reaching the contact page? If few do, check whether the link is easy to see, especially on a phone.
- Are people leaving a long page quickly? Perhaps the beginning does not match what they expected.
- Is one source bringing steady visitors? You might write more on those topics or keep those pages fresh.
- Did a change coincide with a drop? Check whether something broke, such as a form or page. A short test of the site is usually more informative than the chart.
If you want to compare two versions of a page, remember that split testing needs enough visits to be meaningful. On very small sites, listening to a few real people can teach you more than a statistics report.
A simple monthly review
- Choose a fixed period, such as the last full month.
- Note total visits and total page views.
- Look at the top five pages.
- Look at the main sources.
- Write down one thing you noticed and one thing you will do about it, even if the answer is “nothing yet.”
- Keep the notes so you can compare over several months.
Keep it short. A tidy ten-minute habit beats an elaborate dashboard that you never open.
Privacy considerations
Analytics tools may set cookies or collect details about visitors, and some regions have rules about how this must be disclosed or whether consent is needed. Make sure your privacy policy and any consent banner describe what you really use, and collect only what you need. Rules vary, and this is not legal advice.
Used with humility, basic analytics are a helpful lamp for a small site, not a verdict on its worth.