Website analytics – quantitative and qualitative analysis, tools and measurement in the AI era

Definition

Website analytics (web analytics) is the collection, measurement and analysis of data on how visitors find a website and what they do there. The purpose is to understand what works and what does not – so that decisions about content, design and marketing are based on data rather than gut feeling.

What does website analytics include?

In practice, website analytics splits into two tracks that answer different questions. Quantitative analysis shows what is happening on the website. Qualitative analysis shows why it is happening. Most companies stop at the first track and so miss half the picture.

Quantitative web analytics is based on figures from measurement tools and answers questions such as:

• How many visitors come, and from which channels – organic search, ads, email, AI assistants?
• Which pages are visited, for how long and in what order?
• What share of visitors do what the website exists for – a purchase, a quote request, a booking?

Qualitative web analytics looks at behaviour and motives, using heatmaps, session recordings, short on-site surveys and user testing. Quantitative data can show that many people abandon a form halfway – it is the qualitative analysis that shows which field stops them. In practice you need both: the numbers point out where the problem is, and the qualitative material shows what to do about it.

Website analytics is not the same thing as reporting. A report view that nobody acts on is statistics, not analysis. The work consists of three steps: collect the right data, interpret it against a business goal and make a decision – change, test or reprioritise. If the last step is missing, the choice of tool does not matter. That is also why website analytics acts as the hub between the disciplines: SEO, advertising, CRO and AEO are all measured against the same data.

Which methods and metrics are used in website analytics?

Which metrics are relevant depends on the website's goals, but a few come up in almost every analysis:

• Traffic by channel – where visitors come from and how each channel performs over time.
• Conversion rate – for most business websites the single most important metric, because it reflects business, not just traffic.
• CTR – shows how much of your visibility in search results, ads and email campaigns actually turns into visits.
• Engagement – GA4 measures engaged sessions.
• Attribution – decides which channel is credited with a conversion when the customer journey has several touchpoints, and therefore where the budget ends up.

On top of the metrics sit the methods. Segmentation compares groups, for example mobile versus desktop or new versus returning visitors. Funnel analysis shows where in a flow visitors drop out. Cohort analysis follows a group over time. And A/B testing replaces guesses with controlled experiments. Together, this is the raw material for CRO – without reliable measurement there is no way to know whether a change made any difference.

Choose a few metrics with a clear link to the business rather than many without one. An e-commerce store and a B2B website for lead generation need different sets – but both need to be able to answer the same basic question: did what we changed lead to more business?

Which tools do you need for website analytics?

For most companies, a small set is enough, and most of it is free:

• Google Analytics 4 (GA4) – the standard tool for quantitative analysis: channels, behaviour and conversion events.
• Google Search Console – shows how the website performs in Google's search results before the click: impressions, position and CTR per search query. A complement to GA4, not an alternative.
• Looker Studio – Google's free reporting tool that brings GA4, Search Console and ad data together in a shareable report view.
• Google Tag Manager – manages the measurement scripts and makes it easier to control which ones load before and after consent.
• Microsoft Clarity or Hotjar – qualitative tools for heatmaps and session recordings. Clarity is free.
• Matomo, Plausible and other cookie-free alternatives – for organisations that want to reduce GDPR risk or avoid a consent banner. We compare them in Alternatives to Google Analytics.

The choice of tool is rarely the bottleneck. The most common problem is a standard installation with no measurement plan: no conversion events, no link to business goals and reports nobody looks at. A well-thought-out measurement plan in GA4 gives you more than five extra tools would without one.

What does GDPR require of your website analytics?

Two sets of rules govern website analytics in Sweden, and they are often confused. The cookie question itself is regulated by the Swedish Electronic Communications Act (lagen om elektronisk kommunikation, LEK), with the Swedish Post and Telecom Authority (PTS) as the supervisory authority: cookies that are not necessary require consent before they are set. When the tracking also processes personal data – which classic analytics tracking generally does – the EU's GDPR applies in parallel, with the Swedish Authority for Privacy Protection (IMY) as the supervisory authority.

IMY has issued administrative fines of between SEK 8 million and SEK 37 million against Swedish companies for tracking-related GDPR breaches, including unlawful sharing of visitor data with third parties. So the fines concerned the handling of personal data – not the cookie banner itself. In spring 2025, IMY also reviewed how Swedish websites design their consent banners; the review led to criticism but no fines.

In practice, this means three things for your website analytics:

• Do not load any analytics scripts until the visitor has given consent – in a Google set-up this is handled via Consent Mode.
• Do not collect more data than the analysis requires, and switch off features you do not use.
• Consider cookie-free tools if a low consent rate is eroding your data.

How is website analytics changing in the AI era?

More and more customer journeys start in an AI assistant instead of on a search results page. For a long time, that traffic was effectively invisible in the standard reports – it was lumped together with referral and direct traffic. That changed in May 2026.

On 13 May 2026, Google introduced AI Assistant as a channel of its own in GA4's default channel group. Traffic from known AI assistants is automatically tagged with the medium ai-assistant – which assistants are included is set out in Google's channel documentation, a list that is updated regularly and currently includes ChatGPT, Gemini and Copilot, among others. No configuration is needed, but the classification is not retroactive – historical traffic is not regrouped.

There are two important caveats. At the time of writing, Perplexity is not on the list and lands in the Referral channel, and Google's own AI Overviews and AI Mode count as organic search, not as AI traffic. In addition, a significant – but currently not reliably measurable – share of AI traffic has no referrer and ends up in Direct. In other words, the AI Assistant channel captures part of the picture, not all of it.

Is the traffic worth watching? An early data point comes from the US agency Seer Interactive: for a single client, ChatGPT traffic converted at 15.9%, compared with 1.76% for organic search. The sample is small – around 11,000 AI sessions over seven months at a single international company – and Seer itself describes it as a signal rather than a benchmark. But the signal is plausible: someone who has already had their question answered by an AI and still clicks through is often further along in the buying journey.

The bigger shift is that visibility in AI answers does not always produce clicks to measure. Your brand can be mentioned and recommended without a single visit being recorded – that is the side of AEO that classic website analytics cannot see. We go further into how attribution is affected when AI answers take over parts of the customer journey, and how GA4 fits into AEO tracking, in the glossary entries on attribution and Google Analytics.

In 2026, then, website analytics means measuring two things at once: classic traffic and the emerging AI traffic. Want help setting up the measurement – or appearing in the AI answers it measures? Get in touch.