Updated 30 October 2026: Rewritten as a practical review guide. Unsupported promotional and outcome claims were removed. The original publication date remains.
Conversion rate optimisation starts with a specific customer task: requesting a proposal, booking a consultation or completing a purchase. Find where that task becomes difficult, make a focused change and check whether the evidence supports keeping it.
What should count as a conversion?
Choose a completed action that matters to the business, and define its denominator. For example, a completed-enquiry rate might use successful submissions divided by sessions to the relevant service page. State that scope consistently; a button click, a stored enquiry and a qualified sales opportunity are different records.
Exclude documented test submissions from business reporting and review duplicates separately. Consent choices and collection failures can create differences between form storage and analytics. Google’s GA4 funnel guidance explains how entry conditions affect a funnel: an open funnel allows entry at any step, so do not read it as a single compulsory journey.
Which usability problems should you check first?
Check the offer, the next action and the completion experience before adding more interruptions. Can a visitor understand the service, see what information is required, use the form on a phone and receive an accurate confirmation?
- Offer: explain the scope, relevant limits and what happens after contact.
- Controls: use clear labels, visible focus and useful error messages; check keyboard access.
- Form: request information needed for the next step and explain optional fields.
- Completion: verify successful storage and delivery, rather than counting only a click on Submit.
A pop-up or live chat may help a particular task, but neither is a universal improvement. Check whether it obscures content, interrupts mobile use or creates a response expectation the team cannot meet.
How should you choose and assess a test?
Write a hypothesis with a reason, a specific change, a primary outcome and a check for unwanted effects. If traffic is too limited for a reliable comparison, start with observed usability failures and describe the resulting evidence honestly.
Hypothetical example: visitors repeatedly ask which service a general contact form covers. A team could test a clearly labelled service field and a short explanation of the reply process. It should compare completed enquiries and their relevance, while checking form errors and response workload. This is a proposed test, not a reported SEM Lion client result.
Does a higher conversion rate mean better business results?
No. More submissions can include duplicates, spam or poor-fit requests. Review qualification, completed sales and the cost of handling enquiries where that evidence exists. A small observed change also needs context: traffic mix, seasonality and measurement changes can affect the result.
A practical prompt to adapt
“Review this customer journey using [page URL], [intended customer action], [form fields], [observed usability problems] and [aggregate measurement data]. Identify three testable hypotheses. For each, state the evidence, proposed change, primary outcome, unwanted effects to monitor and what remains uncertain. Do not invent conversion results or include personal enquiry details.”
Before testing a new page or form, our conversion measurement audit can help establish the baseline. SEM Lion’s conversion optimisation service can then be scoped around the customer journey, available evidence, approval process and acceptance checks.