Diagnose the landing page before you redesign it
Sometimes the page needs a rebuild. Sometimes it needs one working button and a more honest promise.
A disappointing conversion rate is a symptom, not a design brief. Replacing the page before understanding the symptom can produce a more expensive version of the same problem. A beautiful page that answers the wrong question is still answering the wrong question—just with nicer spacing.
Start by asking what visitors expected, what they encountered and where progress stopped. The goal is to identify the smallest change that addresses the strongest evidence. A redesign remains an option. It simply has to earn its place in the queue.
Check that the problem exists
Imagine a hypothetical B2B service page whose reported conversion rate drops after a release. The immediate request is to simplify the form. But the form still works and qualified enquiries in the CRM are stable. The analytics completion event has stopped firing on one route. Redesigning the page would not fix the reporting failure.
Reconcile the event count with actual enquiries. Test desktop and mobile, different form paths, validation failures and success states. Confirm that the denominator is comparable: users, sessions and clicks are different measures. If a traffic source changed at the same time, an aggregate conversion rate can move even when each audience behaves similarly.
Record the suspected problem in one sentence. “Mobile visitors cannot complete the company-size field” is actionable. “The page does not feel premium” may be useful creative feedback, but it is not yet evidence of a conversion obstacle.
Audit the promise across the journey
Read the query or audience context, the ad and the first screen as a continuous conversation. Does the page deliver the promised service, geography and next step? If an ad invites a price comparison and the page offers only a generic company biography, visitors have to do the translation themselves.
Look for concealed eligibility rules. A service may only support certain business types or locations. Making that clear can reduce raw enquiries while increasing useful ones. A lower form conversion rate can be a reasonable trade if unsuitable demand is filtered earlier and suitable visitors still understand the offer.
Ask a person unfamiliar with the page to explain what it offers after a short look. Do not coach them first. If their answer differs from the intended offer, the message needs work before the decorative layer does.
Inspect four kinds of friction
Technical friction includes broken fields, slow essential content, inaccessible controls and lost form entries. Fix reproducible defects before running an experiment that knowingly sends people to a broken journey. This is maintenance, not a thrilling showdown between Button A and Button B.
Cognitive friction is the work required to understand the offer. Conflicting headings, unexplained terminology and several equally loud calls to action can make a simple choice feel complicated. Use a clear sequence: what it is, whom it helps, why the claim is credible and what happens next.
Trust friction appears when visitors cannot judge the claim or the cost of responding. Specific service details, relevant proof and honest next-step expectations are more useful than unverified badges. Never invent a testimonial because the layout has a convenient testimonial-shaped hole.
Operational friction happens after submission. A form that leads to a silent inbox is not a complete conversion journey. Check routing, acknowledgement, ownership and the promised response process. The landing page may be innocent; the handoff may be eating the results.
Prioritise by evidence and consequence
Use a practical order. P0: failures that prevent a legitimate enquiry or create misleading records. P1: strong mismatches between visitor intent and the offer. P2: plausible improvements supported by behavioural or sales evidence. P3: cosmetic ideas with no demonstrated impact yet.
For each proposed change, write the observation, hypothesis, affected audience and success measure. “Clarifying eligibility will reduce unsuitable enquiries without materially reducing suitable enquiries” is testable. “Make it modern” is a direction for design exploration, not a business hypothesis.
Keep a record of the release and avoid changing traffic, offer and form simultaneously unless there is a compelling operational reason. If changes must happen together, report the combined outcome without claiming to know which component caused it.
Read the result beyond the form
A shorter form can increase submissions while removing information sales needs. A longer form can reduce spam while discouraging suitable people. Neither direction is universally correct. Compare completion, qualification, follow-up effort and later outcomes using reasonably mature cohorts.
When traffic is too limited for a credible comparative test, combine structured usability review, technical checks and staged releases. Call that evidence what it is. Do not dress a handful of observations as statistical certainty.
Friction is a useful physics analogy because it draws attention to resistance. Its limit is equally useful: people are not objects on a ramp, and adding a qualifying question can improve the system even while increasing effort. Remove pointless work; preserve necessary decisions. The best next step may be a redesign, but now it will have a reason.