How to measure whether your website is doing its job

Traffic, engagement and search visibility can help diagnose a website. They cannot answer the commercial question on their own: is the site helping the right people understand, evaluate and work with the firm?

There is no useful universal conversion rate for every B2B service website. A firm with a small number of high-value buying journeys needs a different measure from a membership body, lender or high-volume advisory service. Start with the website's jobs and build evidence around them.

Define the jobs before the metrics

A professional-services website may need to:

• help a referred prospect verify expertise
• make a specialist capability discoverable
• give a buying group evidence to circulate
• generate and qualify appropriate enquiries
• support recruitment or media due diligence
• let existing clients complete routine tasks
• distribute useful thinking to a defined audience

Prioritise the jobs and connect each to a user, task and organisational outcome. “Raise awareness” needs a more observable definition, such as increasing recognition among a named audience or helping target buyers find a particular proposition.

One site can have several jobs. Do not force every page to generate an enquiry.

Measure enquiries through the whole service

Count enquiries, then classify them consistently. Define what makes an enquiry relevant, qualified, progressed and instructed. Record unsuitable or harmful contacts too; a higher total can create more work without more value.

Trace the journey beyond form submission:

• successful submission and error
• time to first meaningful response
• route and owner
• qualification outcome
• progression, instruction and reason for loss

This reveals whether the website or the follow-up process is limiting performance.

Segment by source and landing journey where sample size allows. Avoid comparing small groups as if minor changes were meaningful. Keep definitions stable and report uncertainty.

Inspect the content that supports decisions

Top pages by traffic often reflect broad search demand rather than commercial contribution. Examine pages used in relevant journeys: services, sectors, cases, people, tools and authoritative articles.

Useful signals include:

• entrances from target queries or referrals
• movement to supporting evidence
• use before an enquiry or known conversation
• return visits to a topic over time
• downloads or tool completions with appropriate consent
• direct feedback from prospects and client teams

Analytics paths do not show what a person thought or which page caused a decision. A case page appearing before an enquiry is evidence of exposure, not proof of influence.

Review old high-traffic material for currency and fit. An obsolete regulatory article can attract attention while damaging trust.

Use return behaviour with context

Complex buying decisions often involve several visits and people, but a return-visitor rate is an imperfect proxy. Consent choices, devices, browsers and analytics settings can fragment the same person. Several people from one organisation may appear as unrelated users.

Use return patterns to generate questions. Which propositions bring people back? Do later visits move towards evidence or contact? Are known target accounts engaging where lawful and appropriately governed?

A low return rate may be healthy for a page that answers one question completely. Interpret the measure by task.

Ask people how the website contributed

Add a consistent attribution question to the earliest appropriate conversation: “How did you first hear about us?” Follow with “What did you use when deciding whether to contact us?”

The distinction matters. A referral may create awareness and the website may provide verification. A conference may begin the relationship and an article may sustain it. First-touch and influence are different.

Use a controlled set of categories with free text and an “unknown” option. Train colleagues to ask neutrally and record the answer in one place. Do not force the website into one source simply to simplify reporting.

Self-reported memory is incomplete. Combine it with consented analytics, CRM evidence and qualitative feedback rather than treating it as ground truth.

Add the qualitative layer

Speak to recent clients and suitable prospects. Ask what they needed to know, whether they visited, what increased or reduced confidence and what they could not find.

Lost-pitch debriefs can include the digital experience without assuming it affected the result. Internal teams also know which links they share, which pages they avoid and what buyers request repeatedly.

Use a consistent discussion guide and record themes. Protect confidentiality and obtain permission before publishing a remark. A vivid comment is valuable evidence of one experience, not a sector statistic.

Task observation adds another layer. Watch representative users locate expertise, compare evidence or make an enquiry. This can explain an analytics drop-off faster than a dashboard.

Build a board view that supports decisions

A one-page monthly or quarterly view should reflect the site's jobs. Include:

• qualified demand and progression
• performance of priority journeys
• content or evidence supporting those journeys
• service failures and response
• relevant qualitative findings
• changes made and what happened afterwards
• decisions or investment required

Show trends with counts and definitions. Avoid percentages without denominators. Mark low-volume results and known data gaps.

Connect activity to commercial outcomes without claiming exclusive credit. “Three instructed matters first heard about us through referral and used these case pages during evaluation” is more credible than assigning all revenue to the last tracked click.

If you want to build this for your own website, we've put together a template you can adapt - it's the same framework we use when we start working with a new client.

Establish a baseline before optimisation

Document current tracking, consent, data quality and hand-offs. Test forms and CRM capture. Agree who owns classification and follow-up. Then establish several periods of baseline data where volume permits.

When making a change, state the hypothesis and primary measure. Include guardrails for enquiry quality, accessibility, privacy and operational effort. Seasonality, campaigns and market changes may affect the result, so describe causal confidence accurately.

The companion article on small UX fixes and business impact explains how to prioritise interventions. The broader discussion of why B2B service websites fail users helps diagnose structural problems.

A website is doing its job when evidence shows it helps priority audiences complete useful decisions and tasks, while the firm can operate the resulting service well. Measurement should make that contribution clearer, including where the site is being given credit it has not earned.