The business case for a rebuild: how to tell a real one from an expensive redecoration

Most redesign proposals are argued on the wrong grounds. "It looks dated" is true and is not a business case, because it does not predict anything — and the people who will approve the budget know that.

Here is how to tell whether a rebuild will actually move something, and how to say so in terms that survive a finance conversation.

Four questions, in order

One: where are people actually leaving?

Before any design work, find the step that loses the most people. It is often not the one that looks worst.

Where they leave	Usually means	Fix
Before the page renders	Slow first paint on a real connection	Performance, not design
On the first screen	The page does not say what this is	Copy and hierarchy
Mid-page	The next question is not answered where it is asked	Content sequence
At the form	Too many fields, or an error they cannot resolve	Form design and validation
After contact	Response time or qualification	Not a website problem at all

Two: does the site load on the connection your buyers actually use?

Not on the agency's laptop. Test on a mid-range Android on a throttled connection, which is what a large share of the world browses on.

Three numbers, measured before and after, on the same conditions:

1. Largest Contentful Paint

   When the main content appears. This is the one that correlates most directly with people leaving before they have seen anything.

2. Interaction to Next Paint

   Whether the site responds when tapped. A page that renders fast and then ignores taps for 300ms reads as broken.

3. Cumulative Layout Shift

   Whether things move under a finger. The cause of a great many accidental taps and a great deal of mistrust.

These have a business case that does not require a designer's opinion: they are measurable before, measurable after, and they move for reasons you can point at.

Three: can somebody who has never heard of you tell what you do?

The test is mechanical. Show the first screen to five people outside your industry for five seconds, then ask what the company does and who for.

If they cannot say, the problem is copy and hierarchy — not colour, not typography, not animation. A redesign that changes the aesthetic and keeps the copy will produce a better-looking page that answers the same question badly.

Four: does it work for everyone who might buy?

Accessibility is usually argued on ethics and compliance, both valid. The commercial argument is narrower and harder to dismiss: the fixes are the same fixes that make a site work properly for everybody.

1. Contrast failures affect anyone in daylight

   Not a minority case. A phone in sunlight has the same problem as low vision.

2. Touch targets under 44px are missed by everybody

   Especially on a moving train, one-handed.

3. Forms without proper labels break autofill

   Which adds friction for every user at the exact moment they were about to convert.

4. A keyboard-navigable site is a testable site

   The same structure that lets a screen reader work lets an automated test drive the form.

What to commit to before starting

A redesign with no stated metric cannot succeed or fail, which is how it becomes a matter of taste — and taste arguments are decided by whoever is most senior.

1. Baseline the funnel and the three performance numbers

   Same conditions, written down, before anything changes.

2. Name the one step you are trying to move, and by how much

   "Reduce drop-off at the enquiry form" is a project. "Modernise the brand" is not.

3. Agree what you will not change

   Usually the URL structure and the copy that already works. Unbounded scope is what turns a six-week project into six months.

4. Plan to measure four weeks after launch, not four days

   Novelty and re-indexing both distort the first fortnight.

The failure mode: prettier and worse

It happens often enough to name. A redesign ships, the team is pleased, and conversion drops. Four causes account for most cases:

1. More weight

   A hero video, a font family with six weights, an animation library. Every one adds load time, and load time was possibly the original problem.

2. Cleverness over clarity

   A scroll-jacked narrative is impressive once and obstructive every time after. The visitor who has already decided now cannot find the contact link.

3. Copy shortened to fit a layout

   The paragraph that answered the buyer's real objection is cut because it did not fit the grid. This is the most damaging and the least noticed.

4. URLs changed without redirects

   A year of accumulated search ranking discarded in one deploy.

When the answer is no

Sometimes the honest advice is not to rebuild:

1. The problem is response time, not the site

   If enquiries arrive and are answered in three days, a new site produces more enquiries that are answered in three days.

2. Traffic is too low to attribute anything

   Below a few hundred conversions a month, you cannot tell a real improvement from noise. Acquisition first.

3. The site is slow for one fixable reason

   Fix that and re-measure. It is a fraction of the cost and it might be the whole problem.

4. Nobody can say which metric should move

   Then the project has no success condition, and it will be judged on whether people like the colours.

The rebuild worth doing is the one where you can say, before starting, which number you expect to move and how you will know. Everything else is redecoration — which is sometimes the right call, and should be described honestly as what it is.