Web analytics

What it means

Web analytics measures where a visitor came from, what they did on the site and whether they reached an order or a form. It is not a report but an estimate assembled by scripts running in the visitor’s own browser.

Its figures never match the revenue in the books exactly: script blockers, cancelled orders and returns move the result both ways. Without cookie consent part of the traffic is not measured at all, and GDPR limits which identifiers may be stored.

Web analytics is not business intelligence. Analytics sees only what happens on the site, from arrival to the order being sent. What happens next, whether the customer paid, returned the goods or bought again, is known only to accounting and the CRM; only combining these sources shows which customer actually makes the company money.

How it works in practice

In an online shop the goal is set as a paid order, not a view of the thank-you page. Otherwise the cost of acquiring a customer comes out lower than it really is, and the advertising budget is planned on an invented number.

Analytics is processing of personal data: it belongs in the record of processing activities, needs a processor agreement and a defined retention period. The Slovak data protection authority supervises it.

The company sends a newsletter and also pays for ads, but in analytics both kinds of visit merge into one source. Campaign links therefore get UTM parameters under one consistent scheme: source, medium and campaign name. Only then can a campaign’s result be set against what it cost.

What to watch out for

Conversions must not be added up across ad accounts: Google and Meta both claim the same order, so the company sees twice the volume. It then spends months funding a channel that earns less than it costs; the only reliable source is the number of paid orders in the books.

In a website redesign the page addresses change and the thank-you page the goal was tied to disappears. Tracking carries on, the dashboard shows conversions collapsing and the company spends months looking for a fault in its ads, when measurement simply stopped. After every change to the site the goals should be checked with a test order.