MarketingFramework / modelModelAccessible

Web analytics

How can web analytics support strategic choice or positioning?

AccessibleStrategicTeam3 min read

Contents

Web analytics is the process of analysing online behaviour so as to optimise website use and increase engagement and sales.

Web analytics measures how people discover, use and respond to digital properties so an organisation can improve experience, engagement and commercial outcomes. It combines market context with behavioural data from the organisation’s own sites and experiments.

When to use it

Review off-site market and competitor signals periodically—annually in a stable market and more often when conditions move quickly. Monitor on-site performance continuously enough to detect technical failures, campaign effects and changing journeys.

Use the channel to test messages, offers, pricing and landing experiences before committing to more expensive media, while recognising that behaviour may differ between online tests and other contexts.

Questions include:

How many eligible people visit, and how is that changing?

Which audiences and acquisition sources create useful engagement or value?

What searches, pages and journeys lead to success or abandonment?

Which content is unused, confusing or technically weak?

How long and how deeply do people engage, using metrics appropriate to the experience?

What proportion completes the intended outcome?

Which market and competitor trends should influence the digital strategy?

Origins

Web analytics began with server-log analysis in the early commercial web and expanded through page tagging, cookies, campaign attribution and digital experimentation. As sites became applications used across devices, the field moved from counting page views toward events, users, journeys and business outcomes. Contemporary practice must also account for consent, browser restrictions, data minimisation and the limits of cross-device identity.

What it is

Off-site analytics examines the wider digital market: potential audience, search demand, competitors, referral ecosystems and trends. On-site analytics measures behaviour within properties you control, including acquisition, navigation, events, conversion and retention.

Why it matters

A website’s existence does not prove that it helps users or the business. Analytics makes performance observable and allows teams to diagnose weak journeys, prioritise research and test changes quickly.

Digital experiments can compare landing pages, messages or offers before broader rollout. Speed and low marginal distribution cost are advantages, but valid testing still requires a clear hypothesis, appropriate randomisation, enough observations, guardrail metrics and protection against repeated peeking.

How to use it

Start with business and user questions, then create a measurement plan linking each question to an event, property, denominator and decision. Implement collection with consent and privacy controls, validate it against real journeys and maintain a data dictionary.

Use suitable tools rather than building commodity collection from scratch, but do not assume default reports understand your business. Segment traffic, examine funnels and cohorts, investigate qualitative context and connect web outcomes with downstream value where lawful and reliable.

Heat maps and interaction tools can show where people click or stop, but “hot” areas are not automatically good and “cold” areas are not automatically unnecessary. Interpret them against page purpose, accessibility, device and user intent.

Practical example

Yahoo! used high-volume web traffic to run controlled experiments on its home page. Users were randomly assigned to variants or a control, allowing the company to estimate whether a change caused the desired behaviour rather than merely coinciding with it.

Results could appear quickly and inform broader rollout. The organisation reportedly ran about 20 experiments at a time, replacing opinion-led design debates with evidence while retaining the need to watch long-term and unintended effects.

Top practical tip

Maintain a small scorecard tied to user and business outcomes, annotate releases and campaigns, and review trends with qualitative evidence. Use experiments for causal questions and observational analytics for diagnosis.

Top pitfall

Do not optimise easy proxy metrics while harming trust, accessibility, retention or profit. Validate tracking, respect consent, avoid dark patterns and distinguish correlation from the causal effect of a change.