The problem with small-site analytics
It is not lack of data. It is that a tool built for large sites presents a small site with dozens of metrics, most of which are statistically meaningless at low volume, and the owner ends up either checking obsessively or never at all.
With a few hundred visitors a month, a conversion rate that moves from 2% to 3% is two extra people. That is not a trend, it is Tuesday.
The five numbers
1. Visitors, monthly, compared to last month and to the same month last year. The year-over-year comparison matters because most sites are seasonal and month-over-month noise hides it.
2. Top ten pages. Where attention actually goes, which is almost never where you assumed. Small sites routinely discover that one old post carries half their traffic.
3. Traffic sources. Search, direct, referral, social. The shape tells you which of your efforts is producing anything.
4. Search queries from Search Console. The exact words people used before clicking. This is the single most actionable free data on the internet, and most small sites never open it.
5. Your one conversion. Signups, orders, enquiries. One number, tracked properly, beats ten tracked loosely.
What to ignore, with permission
Bounce rate. On a small content site it mostly means someone read the page and left satisfied.
Time on page. Measured badly by almost every tool, and dominated by open tabs nobody is reading.
Real-time visitors. Compulsive, entertaining, actionable never.
Everything with a decimal point at low volume. Small numbers plus percentages equals false precision.
The monthly review, in fifteen minutes
1. Visitors this month against last month and last year. Note the direction.
2. Top ten pages. Anything new? Anything that dropped?
3. Sources. Did a channel change materially?
4. Search Console: queries where you rank between positions five and fifteen. These are the pages worth improving — you are already close.
5. Conversions. Did what you shipped last month move anything?
6. Write down one thing you will do this month. One.
That is the whole discipline. Doing it every month for a year beats an elaborate dashboard nobody opens.
Choosing a tool at this size
Requirements are modest and specific: it should load in a few kilobytes, be readable without training, not need a consent banner, keep at least a year of history so year-over-year works, and not price you out as you grow.
You do not need funnels, cohorts, session recording, heatmaps or attribution modelling. When you have enough traffic for those to mean something, you will know, because you will have a specific question that the five numbers cannot answer.
The point of the exercise
Analytics is not a scoreboard, it is an input to decisions. If a metric has never changed anything you did, stop collecting it and stop looking at it. What is left will be small, useful, and something you will actually keep up with.