Website Analytics Without the Creepy Tracking

Sometime in the last decade, "install Google Analytics" became step one of every website launch checklist, and a strange bargain came with it: to find out whether anyone read your pricing page, you agreed to help an advertising company follow your visitors around, and you bolted a consent banner onto your site to make that legal. Under EU cookie rules, analytics cookies are not "strictly necessary," so they require consent, which is why the banner cannot simply be removed while the tracker stays.

Here is the part the checklist never mentions: a small business does not need any of that. Every question you actually have about your website can be answered by counting visits in aggregate, without following a single person. The five steps below replace the creepy default with numbers you will actually read.

Step 1: List your questions before you pick a tool

Most analytics regret starts with choosing software first. Flip it. Write down what you want to know, and be suspicious of anything that does not fit on one hand:

1. Is anyone visiting, and is it more than last month?

2. Which pages do they actually read?

3. Where do they come from: search, social, another site, a directory?

4. What did they type into Google to find me?

5. Do they do the thing the site exists for: call, book, buy, fill the form?

Every one of those is a question about totals, not about any individual. That observation does the tool-choosing for you in Step 3.

How to check it worked: next to each question, write the decision it feeds. "Which pages do they read" decides what you write more of. "Where do they come from" decides where your effort goes. If a metric feeds no decision, it does not make the list, however impressive it looks in a dashboard.

Step 2: Take the free search data first

Question 4 does not even need code on your site. Google Search Console is free, and it measures your pages' performance inside Google's results, so it collects nothing from the people visiting your site.

It is also brutally honest. Our own report for the last 28 days shows this blog was displayed 758 times in search results and clicked 6 times. One post was shown 209 times without earning a single click, which tells us its title is losing the argument on the results page and needs a rewrite. That is a real decision, fed by a free report, at zero cost to anyone's privacy.

How to check it worked: verify your site, wait a few days (the data runs about two days behind), then open the Performance report. When your pages start appearing, you have question 4 answered forever.

Step 3: Pick a counter, not a tracker

For questions 1, 2, 3 and 5 you need a counting tool on the site, and this is where the camps split. GA4 is built on a person-level model: it assigns identifiers, stores cookies, and can follow a visitor across sessions, which is exactly the machinery the consent banner exists for. The cookieless tools count the visit and immediately forget the visitor:

Here is what a cookieless dashboard looks like in practice. This is Plausible's public live demo, which anyone can open:

Be honest about the trade. Going cookieless gives up demographic guesses, cross-device journey stitching, and remarketing audiences. If you were not using those, and a site with a few hundred visits a month has no statistical business using them, you are giving up nothing and getting your visitors' trust back.

How to check it worked: you have picked one tool, and you can say out loud why: free and fine with the banner (GA4), simplest paid option with the banner gone (Plausible or Fathom), or free and cookieless with fewer features (Cloudflare Web Analytics).

Step 4: Install it and watch yourself arrive

Every tool in the table installs the same way: one small script tag in your site's head. On WordPress that usually means a plugin or a theme settings box; on most builders it is a field called header code. On StoryPress you ask Claude to add the snippet in plain English and it is on every page at the next publish, which is the kind of chore the $5 a month is supposed to make disappear.

How to check it worked: open your site on your phone, then open the analytics dashboard. Cookieless tools update in near real time, so you should watch yourself get counted within a minute. If nothing appears, the snippet is not in the head of every page, which is the only thing that ever goes wrong here.

Step 5: Wire up the one action that pays

A visit count alone cannot answer question 5, so define one to three events for the actions that pay your bills: the contact form submission, the tap on your phone number, the click to your booking page. Plausible and Fathom both do this without code for common cases like outbound clicks and form completions; GA4 can too, with more setup.

Then put a 15-minute monthly appointment in your calendar: compare visits, top pages, sources and events to last month, and pick exactly one action. That ritual is the entire discipline of small business analytics.

How to check it worked: trigger the event yourself once and watch the counter move. If your own test registers, a customer's real one will.

The afternoon version

• Write the five questions and the decision each one feeds.

• Connect Search Console. Free, no code on your site, nothing collected from visitors.

• Pick one counter from the table. If the banner has been bothering you, pick a cookieless one.

• Paste the snippet, visit your own site, watch yourself get counted.

• Set one event for the action that pays you, and book 15 minutes a month to look.

Measured this way, your analytics answers five questions, respects every visitor, and may let you remove the most hated element on your website. Not bad for an afternoon.