How do you know your new website is ready to launch?
Your new website should help customers find the information they need and finish important tasks. Compare it with the old site, test the full result and record what you are approving before launch.

Before approving your replacement website, agree what customers must still be able to find and do. Compare the relevant old and new pages, try those tasks on the working replacement, and record the result. Investigate differences before deciding whether to approve them: a new heading may be welcome, while an enquiry that never reaches your team needs fixing.
Keep the old website available for comparison while the replacement is on a local computer or staging address. Staging is a working copy used before launch. Ask for both the page-comparison findings and evidence that important customer tasks reach their intended outcome.
Grace WebCompare compares old and new websites, including live, local and staging versions. Use its recorded page fields to find differences for review, alongside hands-on checks of customer tasks.
What should customers be able to finish?
Choose tasks that represent why people use your website. For a service business, one might be: “Find out whether this service suits my situation and send an enquiry.” For a shop, it might be finding a suitable product and completing an order. Use tasks your website actually supports.
Write down what a successful result means before testing. For the enquiry, the customer needs enough information to choose the right service, understand what to provide, send the form and receive a useful confirmation. Your team must receive the enquiry in the agreed place, with the details needed to respond.
Include a task that is easy to overlook, such as finding an older guide or downloading a document linked from a service page. Choose examples across different page types and important paths. A sample gives evidence about the tasks tried; it cannot prove every possible route works.
Did the information people need survive?
Follow each selected task through the old website and note the information that helps someone complete it. Check that information on the replacement, including relevant images, documents and links. A shorter page may explain the service better, but removing eligibility details could leave customers unable to decide whether to enquire.
Check meaning as well as presence. A heading can survive while the useful explanation beneath it disappears. Agree deliberate rewrites and removals with the person responsible for the content rather than assuming that every old sentence must stay.
The old website is a reference, and may contain mistakes of its own. Record what should be retained, improved or removed so the comparison has an agreed purpose.
What can an automated comparison tell you?
Grace WebCompare can compare HTTP status codes, redirect destinations, page titles, meta descriptions, H1 headings and canonical paths between the two sites. These recorded values help identify pages and differences to investigate. It can also check for the presence of JSON-LD, a format for structured page data; presence alone does not establish that the data is correct.
A changed title may be an approved rewrite. A different redirect destination may be intentional or may send someone to the wrong service. Give each important difference a decision and a reason. After a correction, an individual page rescan can check the recorded fields again.
Those findings do not establish that a whole article, image or download survived, that the layout works, or that a form delivers its message. Review the actual content and complete the relevant task. If page addresses change, include a separate plan for old links and search visibility. Our guide to preserving useful search traffic during a redesign covers that work in more detail, alongside Google’s site-move guidance.
Does the task work all the way through?
Try the enquiry on the working replacement. Leave a required field empty and check whether the explanation helps you correct it. Submit valid test details, check the confirmation, and ask the receiving team to find the enquiry. A success message on the screen is only one part of that result.
If your website has accounts or checkout, include the relevant login, permission and order outcomes. Agree safe test arrangements with the developer for anything that could take payment, create a real order or contact a customer. Record which outcomes were tested and which still need confirmation.
Ask a likely customer to try an important task without telling them which menu to open or which button to press. Watch where they hesitate and ask what they expected. GOV.UK’s usability-testing guidance explains this approach. For a deeper guide to observing tasks and reviewing the results, see what to test before approving an app redesign.
Include the browsers and devices relevant to your customers. Check the selected tasks on a phone, and include keyboard use and appropriate accessibility testing in the agreed scope. Mobile layout, usability and accessibility need their own review; a matching set of page fields cannot establish them. GOV.UK’s guidance for running live services similarly treats wider journeys, browser and device testing, accessibility and quality assurance as continuing work.
Approve a customer task from its observed outcome. A page that loads and a form that says “sent” still leave questions to check.
What should the approval record say?
Keep a short record for each selected task: the starting page, the goal, the information that must remain, the expected outcome, and what actually happened. Add the test date and relevant browser or device. Link the comparison finding, screenshot or test record that explains a problem.
Then record the decision. An intentional difference can be accepted with a reason. A missing explanation can go to the content owner for a decision. An essential enquiry that fails needs an owner, a correction and a successful retest before that task is approved. Agree what prevents launch rather than using an unexplained score or treating all differences as equally serious.
For example, the enquiry task might show that the service page is clearer and the form accepts the right details, but the message reaches an unattended inbox. Record the destination problem, assign the fix, then repeat the submission and receipt check. Approval follows the complete result.
Ask the person authorizing launch to review the accepted changes and unresolved items together. Recheck affected work after corrections, and confirm important tasks on the production website when it goes live. Testing on staging provides useful evidence, while the public website has its own configuration and connections.
Questions before you approve
Can an automated website comparison test our enquiry form?
It can identify relevant page and URL differences, but that does not prove the form works. Submit safe test details, check validation and confirmation, and verify that the intended team receives the enquiry.
Should the replacement match every old page exactly?
Agree which information and tasks must remain. Rewrites, navigation changes and removals can be intentional improvements. Record why important differences are accepted and investigate unexplained ones.
How do we choose representative tasks?
Start with what customers need to accomplish, such as choosing a service and sending an enquiry. Include important page types and an easily missed path, such as an older guide or download. State the expected result and remember that a sample does not cover every route.
When is a task ready to approve?
When the observed result meets the agreed expectation, necessary content is available, and relevant findings have been resolved or explicitly accepted. Record remaining limitations, retest corrections and confirm important tasks after launch.