Most redesign cases cannot be checked
The typical business case projects a conversion-rate improvement, multiplies it by current traffic and produces a number. It is rarely revisited, because nobody wrote down what the baseline was, and by the time anyone asks, the comparison is no longer possible.
A case that cannot be falsified is not a case. It is a justification produced after the decision.
Record the baseline before anything changes
Conversion rate by device and by landing page. Qualified enquiries per month, using your definition of qualified. Top twenty pages by clicks with their positions. Average time from enquiry to opportunity, if your CRM knows.
Take this whether or not you proceed, and keep it independent of any supplier. It is the only thing that makes a later claim about improvement verifiable — by you or by us.
Three inputs carry the model
- Value of an enquiry. Close rate multiplied by average deal value. Most organisations can produce this within an hour and most have never written it down.
- Current volume, segmented by device. Aggregates hide the mobile failure that is often the largest single opportunity.
- A defensible improvement assumption. Not a number from a case study on someone else's site.
The third is where cases inflate. If you cannot justify the improvement figure from your own data or a specific diagnosed problem, model a small one. A case that works at a modest improvement is a case; one that needs an optimistic figure is a hope.
Cost the whole thing
Build cost is the visible part. Add content production, internal time in review cycles, the migration work that protects your rankings, and the maintenance obligation that starts at launch. Internal time is the line most often omitted and frequently the second largest.
When the answer is not to proceed
If the model only works with an optimistic assumption, or the site's problem is a message that could be rewritten next week for nothing, or nobody can produce the value of an enquiry, the redesign is not yet justified. Those are useful findings, and they cost nothing to reach.