Analytics Setup That Answers Real Questions

Start from the questions

Before configuring anything, write down what you actually need to know. Which marketing spend produces customers? Where do people abandon the purchase? Which pages bring people in and which lose them?

Configuration then follows the questions. Without them you will collect a great deal of data and answer nothing.

Conversions are the foundation

Define what counts as success: a purchase, an enquiry form, a phone call, a booking, a download.

Without configured conversions, traffic figures are meaningless — a thousand visitors who do nothing is worse than fifty who buy.

Track micro-conversions too, such as reaching a pricing page or starting a form. These show where interest builds and where it is lost.

Attribution, honestly

Customers rarely convert on first contact. They see a social post, search later, click an ad, then return directly a week afterwards.

Last-click attribution credits only the final step and systematically undervalues everything earlier, which leads businesses to cut the activity that created the demand.

No attribution model is correct. Understanding the limitation prevents confident decisions on misleading data.

UTM parameters

Tag campaign links consistently so traffic sources are identifiable. Without them, everything collapses into direct and referral and you cannot tell what worked.

Agree a naming convention and stick to it, since inconsistent capitalisation and naming fragment the reporting.

Privacy and consent

Analytics involving cookies or personal data generally requires consent under privacy law in many jurisdictions, and consent banners must genuinely allow refusal.

This means analytics data is now incomplete by design, since a proportion of visitors decline. Server-side and privacy-focused analytics options exist that collect less and require less consent.

Data retention settings and IP anonymisation are worth configuring deliberately rather than left at defaults.

What to look at regularly

A small dashboard answering the questions you defined, reviewed monthly. Conversions by source. Conversion rate by landing page. Drop-off points in the purchase flow.

Ignore vanity metrics. Total pageviews and time on page rarely change any decision.

Compare like periods — this month against the same month last year rather than against last month — because most businesses have seasonal patterns that make sequential comparison misleading.

The most common configuration errors

Tracking code on some pages and not others. Internal traffic not excluded, so your own visits inflate the figures. Cross-domain tracking not configured, so a hosted checkout appears as a separate visit and breaks attribution entirely.