Website Repair & Speed Optimization
Find the cause before replacing working parts
Slow pages, visual breakage, and failed forms can have different causes. The first step is to reproduce the symptom and identify whether it comes from the page, the software, or the environment.
Tell us what a slow or broken site prevents someone from doing
A site that feels slow to look at and a checkout that stops responding deserve different investigations. Identify the page, device, visitor action, and time where you noticed the issue. If it only appears in a certain location, browser, user state, or product, give that context. A useful symptom tells us where to look rather than forcing the work to depend on one score.
Match the repair to the symptom
Use these categories to describe the trouble. Even where problems overlap, the checks they call for can be different.
- Loading
Identify what customers wait to see and whether the delay comes from the page resource, network, or delivery path.
- Interaction
Record the step a visitor cannot complete. Check the input, response, error handling, and handoff.
- Stability
Look for unexpected movement or broken layout. Test the same view on the screen sizes or devices affected.
Start with a measurable explanation
Test a representative page or task and separate likely causes. A large hero media file, an unexpected layout shift, and slow server response may all feel like lag, though the remedies differ. For a form or interactive feature, record the exact step that fails and what the visitor sees. Match a tested change to that problem, then repeat the same check.
Choose what can be improved without removing what matters
A project might resize images, defer a script, or change how a critical element loads. Some assets are essential to the offer and need better delivery rather than removal. Test price display, forms, or the purchase path around the repair. A performance improvement is only useful if customers can still understand the page and complete the work you need it to support.
Know what an automated scan can and cannot show
A technical scan can give you a starting point for one public page. It is not a complete inventory of every device, visitor, or workflow. Use any field data when available to understand real visits separately from controlled test results. Bring symptoms that are not covered by the scanner into the conversation rather than relying on a score to tell you what went wrong.
Document the fix so it can be checked again
Agree on which paths will be retested and what should change, such as a visual element no longer shifting or a form confirmation reaching the right mailbox. Measure the same page under comparable conditions. Make a note of the original problem, the change, and whether another part of the site depends on it. Those details are useful if the issue returns later.
Bring a page we can test
Tell us what happens and what you expected instead. If the issue is speed, run MJM’s free page scan first and bring the page and results to the conversation.
Start with a specific page or task
Record the affected address, device, browser, and conditions. Find a repeatable test when possible. Use measurements to identify which part of the loading or interaction path deserves attention.
Look for fixes that preserve the experience
Oversized images, blocking scripts, unstable layout, and server response are different problems. Choose the intervention that fits the evidence instead of removing useful content simply to improve one score.
Compare the same thing after the change
Check the original symptom again and test adjacent workflows. A controlled performance test can provide a consistent comparison. Where available, field data describes what real visitors experienced.
How to read Core Web Vitals
These are published thresholds for the 75th percentile of page views. Between the good and poor bands, the metric needs improvement. Each band uses the scale of its own metric. This is a reference graphic, not a measurement of your website.
- Loading the largest visible element (LCP)
- Good 2.5 seconds or lessPoor More than 4 seconds
- Response to interactions (INP)
- Good 200 milliseconds or lessPoor More than 500 milliseconds
- Unexpected visual movement (CLS)
- Good 0.1 or lessPoor More than 0.25
What to bring
Bring the affected URL, a description of the symptom, and any recent change you can identify. Include the model of device or browser if it happens only under certain conditions.
Common questions
Is one speed score the whole answer?
No. It is one indicator. Check what the visitor actually experiences, how the test was run, and whether the result is in lab or field data.
Does an automated accessibility score prove conformance?
No. Automated tests can find some issues. Full evaluation also requires review of content, interaction, and complete user flows.
Tell us what you want to do
Bring a goal, a website address, or the problem you are trying to solve.
