Skip to content
Standout

guide

How to prove a website redesign to a client

6 min read · updated November 17, 2026

A client cannot judge your code and rarely takes your word that the new site looks better. What they can trust is the same measurement taken before and after, with the old version saved before you touched it. This is how to produce that, and what to leave out of it.

Why two screenshots convince nobody

A before and after picture shows that the site is different. It does not show that it is better. The person who chose the two pictures is the person being paid, and the client's business partner is free to say they liked the old one.

Proof is a measurement that could have gone the other way. The same checks, run on the old version and the new one, with numbers the client can read: the contrast that failed and now passes, the phone layout that overflowed and now holds, a score built from a method that is written down.

There is one catch, and it is where most people lose the proof. Sites are usually improved in place. The moment you deploy, the old version is gone, and no tool can measure a page that no longer exists. The baseline has to be saved before the first change.

What makes a before and after fall apart

  • 01

    Nobody saved the before

    After the deploy you are comparing the new site with your memory of the old one.

    fix Before the first change, save screenshots at phone and desktop width and the measurements for each page, with the date.

  • 02

    The two versions were measured differently

    A different tool, a different width or a different set of pages means the difference you report belongs to the method, not the site.

    fix Same tool, same pages, same widths, both times.

  • 03

    A score nobody can explain

    A number with no method behind it invites one question: says who?

    fix Use a score with a published method that gives the same result for the same page, and show which parts of it moved.

  • 04

    Only the home page

    The client's customers land on the service and contact pages too.

    fix Measure the same list of pages before and after.

  • 05

    Promising results you did not measure

    "This will bring more enquiries" is a forecast. Put next to real measurements, it weakens them.

    fix Keep design quality, which you can measure today, apart from business results, which need weeks of data.

  • 06

    Leaving out what is still open

    A client who finds the gap themselves stops trusting the rest of the report.

    fix List what is still open and what you could not check, in the same report as the wins.

Build the proof, in order

  1. 1

    Save the baseline before you change anything

    Phone and desktop screenshots, the measurements and the score for each page, and the date. This is the one step you cannot do later.

  2. 2

    Agree what better means

    Before the work starts, agree the bar with the client in plain words: readable on a phone, text that passes contrast, no dead links, a target score. Then the result is judged against something you both chose.

  3. 3

    Check as you go

    After each round of fixes, run the same checks again and sort the result three ways: fixed, still open, new. A new fault straight after a fix was usually caused by the fix.

  4. 4

    Measure the finished version the same way

    Same pages, same widths, same tool as the baseline. Resist switching to a kinder tool at the end.

  5. 5

    Put the two side by side

    The score before and after, what moved inside it, and both sets of screenshots. Lead with the number, then the pictures.

  6. 6

    Say what is left

    What is still open and why, and what could not be checked. One short honest list makes everything above it more believable.

  7. 7

    Keep business results for later

    Note the launch date. In a few weeks, compare visits and enquiries for matching periods before and after. That is a second report, and it is the one that shows whether the business gained.

What the client actually reads

A client reads a report to answer three questions: is it better, how do you know, and what is left. Write for those three and stop.

Use plain words. "Text on the booking page was too faint to read on a phone and now passes" is something a business owner can repeat to a partner. A rule name and a selector are not. Keep the technical list for yourself and for whoever maintains the site.

Put your own name on it. A report that says who prepared it and when is a document the client can keep.

Frequently asked

I already changed the site. Can I still make a before and after?
Only if a copy of the old version still exists somewhere: a preview deployment, an old commit you can deploy to a preview address, or an archived snapshot. Measure that. If there is none, say so, and treat today's version as the baseline for the next round of work.
Does a higher design score mean more sales?
No one can promise that from a design measurement. The score covers things that get in a visitor's way, such as text that cannot be read and layouts that break on a phone. Whether the business gained is answered later, with visits and enquiries over matching periods.
What should a client-facing report leave out?
Internal jargon, rule names and the fix list in code terms. Keep the score, what was holding the page back, what changed, what is still open, and the pictures.
Can I put my agency's name on the report?
Yes. With Standout you pass your studio's name and the report and its shareable page read as prepared by you.
How long should I keep the baseline?
At least until the client has signed off, and longer if you will do more work on the site. It costs almost nothing to keep and it cannot be recreated.

more guides