Updated October 9, 2026.
A website needs a redesign when its current structure or implementation makes important requirements difficult to meet at a reasonable cost. It needs a focused refresh when the underlying site works and a smaller change can solve the problem.
Age, a dated color palette or a disappointing analytics number does not settle that decision. Start with what customers and staff cannot do, then compare ways to fix it.
Define the problem before choosing the project
Use a concrete description: “Customers cannot submit the quote form on a phone” is more useful than “The website feels old.” Record the affected page, task, evidence and business consequence.
Then agree what success looks like. A working form, a clearer service page or a publishing process staff can complete without help is testable. “A more modern website” leaves too much open to interpretation.
For this guide, a refresh means targeted changes within the existing site. A redesign means a broader change to content, navigation, layouts or visual design. A rebuild replaces substantial implementation. They can overlap, but a redesign does not always require a new platform.
Which problems suggest a refresh?
| Problem | First option to investigate | When broader work may make sense |
|---|---|---|
| Old photos or inaccurate service details | Update content and imagery | The whole service structure or audience has changed |
| A difficult contact form | Repair fields, labels, validation and delivery | Several customer workflows need a different system |
| A slow page | Measure and address the specific bottleneck | Repeated constraints make targeted improvements impractical |
| Hard-to-use mobile navigation | Repair the menu and affected layouts | Page templates need substantial restructuring throughout the site |
| Staff cannot publish updates | Training, permissions or a better editing template | The existing publishing model cannot meet the team’s needs |
| Unsupported software | Contain immediate risk and assess an upgrade path | Essential components cannot be maintained or replaced economically |
A phased plan can be a complete solution. For example, fixing forms and rewriting the main service pages may meet the business need without replacing a reliable site.
Use evidence that explains the problem
Watch a customer task
Ask someone unfamiliar with the site to find a relevant service, check the service area and make an inquiry. Observe where they hesitate. Test on a phone and with a keyboard as well as a desktop mouse.
If you receive repeated calls asking for information already on the site, record the questions. They can reveal content or navigation problems more directly than an overall traffic figure.
Check measurement before interpreting it
Separate visits, button clicks, delivered inquiries, qualified leads and paying customers. A click on a phone link does not establish that a call connected or a sale occurred. Our GA4 event guide explains how to choose useful actions to measure.
Compare relevant pages, devices and traffic sources over comparable periods. Account for seasonality, advertising changes and tracking changes. A drop after a marketing campaign ends is not automatically a website failure.
In GA4, bounce rate is the percentage of sessions that were not engaged; it is not simply the percentage that viewed one page. Google’s engagement and bounce-rate documentation explains the definition and recommends examining traffic sources and tagging. There is no universal bounce-rate cutoff that proves a rebuild is necessary.
Diagnose performance
PageSpeed Insights distinguishes controlled lab tests from real-user data. A low lab score is a starting point for diagnosis. It does not tell you whether the right fix is a smaller image, less JavaScript, a server change or a rebuild.
Test representative service, article and contact pages. Use real-user results when available and repeat controlled tests under comparable conditions. See our website optimization guide for a practical investigation sequence.
Address urgent failures while planning
A broken checkout, missing inquiry delivery or known security incident needs attention before a long design project is complete. Assign an owner for containment and repair, then decide whether broader replacement is justified.
A redesign alone does not resolve compromised accounts, exposed credentials or a vulnerable outside integration. Include those systems in the response. Likewise, a certificate problem may be a configuration fix rather than a reason to rebuild every page.
Prioritize customers’ ability to reach you. If an essential workflow is unavailable, provide a working alternative while the underlying issue is resolved.
Compare options with the same requirements
Ask for two scoped options where feasible: the smallest useful repair or refresh, and the broader redesign or rebuild. Compare:
- Which customer and staff problems each option solves.
- What content, design, integrations and migration work are included.
- What remains unresolved and why.
- Who tests and accepts the work.
- Launch risk, ongoing support and future editing responsibilities.
- Initial cost and additional recurring costs.
Preserve what works: useful content, recognizable branding, successful customer paths and URLs people already use. A framework preference should not replace evidence about your site.
Estimate value without treating leads as customers
The following is an illustrative scenario, not a forecast or client result. Suppose a proposed change would produce 20 additional qualified inquiries per month. Assume 25% become customers and each extra sale contributes $400 after its variable delivery costs.
| Step | Calculation | Result |
|---|---|---|
| Additional customers | 20 inquiries × 25% close rate | 5 per month |
| Additional contribution | 5 customers × $400 | $2,000 per month |
| Less added recurring website cost | $2,000 − $200 | $1,800 per month |
| Simple payback on a $9,000 project | $9,000 ÷ $1,800 | 5 months |
This simplified payback excludes taxes, financing and ramp-up time. It also assumes the extra inquiries actually occur, the close rate holds and the business has capacity to serve the work. Any additional acquisition or staffing cost must be included. Contribution is not the same as total business profit.
Test a less favorable case: at 8 extra qualified inquiries with the same close rate, contribution is $800, or $600 after the added monthly cost. Simple payback becomes 15 months. With no extra inquiries, there is no sales-based payback in this model.
Do not count both “recovered bounced visitors” and the sales those same visitors generate as separate benefits. Value time savings only when you have a credible plan for how that time will be used. Our website cost-of-problems guide develops the assumptions further.
Plan the launch before approving the scope
A useful redesign plan includes:
- A list of important pages, content and customer tasks to preserve.
- A content owner and approval process.
- A URL inventory, with old-to-new mappings where addresses change.
- Mobile, keyboard, form, booking or checkout tests appropriate to the site.
- A backup, rollback plan and owner for post-launch monitoring.
- A measurement baseline and date to review results.
Google’s site-move guidance covers URL mapping, redirects, sitemaps and monitoring. Avoid redirecting every removed page to an unrelated homepage. Search visibility can fluctuate during a move; preserving URLs where practical reduces the number of things that change.
Choose the project you can explain in terms of solved problems, accepted trade-offs and an affordable support plan. If a focused repair meets those requirements, it is a sound investment. If the constraints run throughout the site, a broader build may be easier to maintain and use.