Diagnose a conversion problem before redesigning

Check that the number means the same thing

Define the event being called a conversion. A button click, submitted form, and conversation are different actions. Check the denominator too: visits, users, and advertising clicks are not interchangeable. Reports can appear to disagree simply because they measure different stages or use different periods, even when each report is technically working.

Then look for changes in the measurement setup. A new form or missing confirmation event can affect the reported number. Run a complete enquiry yourself and check that the expected event appears. Fix a measurement problem before attributing the change to design; otherwise you may spend time improving an experience that was not the source of the drop.

• Write the action behind each reported event.
• Use a consistent source and comparison period.
• Separate contact clicks from enquiries actually received.

Compare the traffic before judging the page

Look at the mix of channels, audiences, and devices alongside total visits. A new campaign may bring people who are earlier in their decision process. Their behaviour can change the overall conversion rate without making the existing page worse. Break the data into useful groups before deciding that the headline or form has stopped working.

Planning example: an Indonesian B2B team adds an awareness campaign and sees its combined enquiry rate fall. The team should compare the new traffic with the existing traffic rather than treating the average as a diagnosis. That comparison still does not prove the cause, but it gives the next investigation a more useful direction.

Find friction you can actually observe

Use the page on the devices your visitors use. Does it continue the message from the campaign? Can a new visitor understand the offer and find the details needed to ask a question? Test the form with valid and invalid entries. Record the steps that fail or create uncertainty rather than relying only on personal design preferences.

Bring sales into the review. Repeated misunderstandings can reveal missing service boundaries or unclear information. Poor-fit enquiries might suggest an audience or offer problem rather than a need for more form fields. Connect those observations to the page, then state which change might address them and why it would help the visitor.

• Record the page, device, and steps for each problem.
• Collect recurring buyer questions without sharing personal data.
• Separate missing information from broken functionality.

Prioritise fixes and design honest tests

For each finding, record the evidence, expected effect, effort, and evaluation method. A broken submission route needs a fix. A clearer offer statement is a hypothesis to assess. Explain the mechanism: which visitor decision should become easier after the change? This makes an unsuccessful result useful because the team knows which assumption it was testing.

With limited traffic, use focused observation and usability checks while recognising their limits. A better number after a change does not automatically establish cause and effect. Note campaign changes and operational factors during the same period. Review enquiry quality as well as volume so the team does not optimise for actions that sales cannot use.