The Website Redesign Checklist: 57 Steps Across 8 Phases

A website redesign checklist covers eight phases: a baseline audit of the site you already have, then the seven pillars of the COMPASS framework — Clear Brand Alignment, Operable by Marketing, Meaningful User Experience, Platform Fit, Accountable Performance, Security by Design, and Sustainable Website System. Work them in that order and the expensive decisions get made before anyone opens a design tool. The full checklist is below, 57 items, free to copy into your own project plan.

Most redesigns that disappoint were not badly designed. They were badly sequenced.

The pattern is consistent enough to predict. A marketing director inherits a site that looks dated, the brief goes out as a design brief, agencies pitch on visual direction, and the project runs from moodboard to build. Nine months later the site is prettier, the bounce rate has not moved, and nobody can say why, because the analytics were configured in the final fortnight by whoever had capacity.

Underneath almost every one of those projects is the same root fault: the site was structured around the organisation chart rather than around what a visitor is trying to accomplish. Navigation mirrors departments. Services are grouped by internal reporting lines. A prospect arrives with a specific problem and hits a wall of corporate hierarchy. Redesigning that structure in a nicer typeface changes nothing.

The second fault is scope discovered late. Positioning gets reopened during design. The CMS turns out not to support the campaign workflow marketing actually runs. Security review lands two weeks before go-live and pushes the date by a month. Every one of those is a decision that could have been made in the first six weeks for a fraction of the cost.

A checklist fixes both faults, because it forces the decisions into the right order.

Three notes before you start.

Work top to bottom, but expect two phases to run in parallel. Platform Fit (phase 4) is usually assessed alongside user experience work in phase 3, because the platform decision depends on what the experience needs to do. Operable by Marketing (phase 2) is decided early and then verified at handover.

Assign an owner to every phase, not to the project. A redesign stalls on unassigned approvals far more often than on build complexity. One name per phase, written down, with the authority to sign off.

Treat the unticked items as your risk register. You will not complete all 57 on a lean project, and that is fine. What matters is that you decided to skip an item rather than discovering at launch that nobody had thought about it.

Each item below is a single action. Some carry a one-line explanation of why they matter.

Eight items. This phase is the reason you will be able to prove the redesign worked.

1. Export twelve months of organic traffic, conversions and top landing pages, and store the export outside the site. Without a pre-redesign baseline, no post-launch number means anything.
2. Crawl and inventory every URL on the current site, including PDFs, campaign landing pages, microsites and subdomains.
3. Tag every page in that inventory keep, merge, rewrite or cut, and record who made each decision.
4. Flag every page that currently earns organic traffic or holds backlinks as do-not-delete.
5. Record current Core Web Vitals (LCP, INP, CLS) for the ten highest-traffic page templates.
6. List every system the site integrates with: CRM, marketing automation, event platforms, payment, single sign-on, chat, data layer.
7. Write down the three business questions the new site must be able to answer with data. Which segment converts, which content influences pipeline, which market is underperforming — pick yours and write them as questions.
8. Name one internal decision-maker per phase, and get their availability confirmed for the months the project runs.

Seven items. Set the strategic foundation before a word of page copy is written.

9. Agree a written positioning statement covering who the site is for, what the business stands for, and why a buyer should choose it.
10. Identify two to three priority audience segments and their jobs to be done, and remove the rest from scope. A site written for everyone converts nobody.
11. Interview four to six internal stakeholders across sales, product and leadership, and record every place their descriptions of the business disagree.
12. Write the single sentence a first-time visitor must understand within five seconds of landing on the homepage.
13. Attach one evidenced proof point to each priority segment: a named client, a verified number, a certification, an accreditation.
14. Review the messaging of three direct competitors and note what you can honestly claim that they cannot.
15. Sign off the narrative in writing before the first wireframe is drawn. Reopening positioning during design is the single most expensive rework in a redesign.

Related: Brand Strategy

Seven items. A site your team cannot run without raising a ticket is a liability, whatever it cost to build.

16. List every content change your team requested in the past twelve months and how long each took to go live.
17. Decide who publishes after launch — in-house marketer, agency retainer, or both — before shortlisting any platform. This decision sets the CMS requirements more than any feature comparison will.
18. Write down the honest technical skill level of the people who will edit the site day to day.
19. Require modular, reusable page templates in the build spec, rather than bespoke one-off page builds that only a developer can change.
20. Specify which content types marketing must be able to create unaided: landing pages, blog posts, case studies, events, gated assets, navigation changes.
21. Put CMS training and a written usage guide into the contracted scope, not the optional extras.
22. Set an acceptance test for handover: a named marketer builds a new campaign landing page unaided within two weeks of go-live.

Eight items. Design for how people navigate, not for how you are structured.

23. Map the journey for each priority segment, from first landing page to enquiry, including the pages they visit in between.
24. Draft the information architecture around what visitors are trying to accomplish rather than around your reporting lines. Fix the architecture and half the other problems disappear.
25. Run a card sort with eight to fifteen real users to validate your proposed navigation labels against their mental models.
26. Tree-test the draft architecture before any design work begins, so navigation failures surface while they are still cheap.
27. Define the single next step every page template offers, and delete competing calls to action.
28. Prototype the homepage and three to five key pages with real messaging in them, not placeholder text. A prototype with the actual narrative running through it is what stakeholders can meaningfully approve.
29. Usability-test that prototype with roughly five users per priority segment before production code is written.
30. Specify accessibility to WCAG 2.2 AA at the design stage, including colour contrast, focus states and keyboard navigation, so it is not retrofitted under launch pressure.

Related: UX Planning

Seven items. Usually runs in parallel with phase 3.

31. Write your platform requirements out of phases 2 and 3 before you evaluate a single vendor.
32. Score each candidate platform on three-year total cost of ownership: licences, hosting, support, upgrades and developer time — not build cost alone.
33. Check every candidate against the integration list from phase 0, and confirm the integrations are supported rather than merely possible.
34. Confirm hosting model, data residency and procurement compliance. For regulated, government and institutional buyers in Singapore and Southeast Asia this frequently decides the shortlist on its own.
35. Check whether you can hire or contract for that platform locally, at a rate you can sustain.
36. Ask each vendor how content and data are exported if you leave. A platform you cannot exit is a platform that prices you later.
37. Document the platform decision and the reasoning behind it. Your successor will inherit this choice and should not have to guess.

Seven items. Measurement gets built in, not bolted on afterwards.

38. Define your conversion events and the segment dimensions you will report on before the build starts.
39. Put technical SEO in the build specification: crawlable navigation, clean heading hierarchy, server-rendered content, XML sitemap, canonical rules, robots directives.
40. Produce a complete 301 redirect map from the phase 0 URL inventory, and test it on staging before launch day.
41. Preserve the exact URL of every page flagged do-not-delete, or redirect it one-to-one to its closest replacement. Redirecting old pages to the homepage discards their value.
42. Configure analytics, consent management and CRM tracking on staging, and verify every conversion event fires before go-live.
43. Add structured data for the entities you own: Organization, Service, Article, FAQ, BreadcrumbList. Clean structured data and self-contained answers are what AI engines quote when they generate an answer about your category.
44. Agree the reporting set for the first two quarters after launch, and who receives it, before the project closes.

Seven items. Security deferred is security compounded.

45. Require HTTPS across the whole site, HSTS enabled, and a modern TLS configuration.
46. Write OWASP-aligned development into the contract rather than assuming it.
47. Book vulnerability assessment and penetration testing (VAPT) before launch, with remediation time inside the timeline, not after it.
48. Review PDPA and GDPR obligations for every form, cookie and third-party script on the new site.
49. Document what each form collects, where that data goes, who can access it, and how long it is retained.
50. Assign patch ownership and cadence in writing, covering operating system, CMS and plugin or module level.
51. Run a formal accessibility audit against WCAG 2.2 AA and fix the failures before launch, not in a later phase that never gets funded.

Six items. This is the phase that ends the rebuild cycle.

52. Contract the warranty period and state exactly what it covers, in writing, before signing.
53. Put quarterly patching, CVE monitoring and monthly reporting on a maintenance agreement that starts at launch, not in next year's budget.
54. Schedule a 90-day post-launch review against the phase 0 baseline numbers, with the same people who signed the brief in the room.
55. Set a content ownership rota: who reviews which sections of the site, and how often.
56. Allocate a fixed monthly budget for continuous improvement. Sites that get small improvements monthly do not need replacing every three years.
57. Diarise an annual re-score across all seven pillars, so decay is caught as a maintenance item rather than discovered as a rebuild.

Related: WebCare & Maintenance

If you are reviewing agency proposals rather than running the redesign process yourself, use the checklist above as a scoring sheet. A credible website redesign proposal answers six things directly:

• Strategy before design. Named deliverables for positioning, audience and information architecture that land before visual design starts.
• A decision gate. A defined point where you can review the thinking and stop, having received something usable, rather than committing the full budget on faith.
• Platform reasoning. Why this CMS, scored on three-year total cost of ownership, against your team's actual capabilities.
• Measurement in scope. Analytics, conversion tracking, redirect mapping and technical SEO as line items, not as assumptions.
• Security and accessibility in the timeline. VAPT, remediation and WCAG work with dates attached, before the launch date.
• What happens after launch. Warranty length, patching cadence, training, and who owns the site on day 91.

A proposal that prices the build but treats the other five as unquantified extras is a proposal that will be re-scoped mid-project.

The eight phases map to COMPASS, Construct Digital's seven-pillar Brand and Build framework for strategy-led website rebuilds. Each pillar exists because it corresponds to one of the seven most expensive mistakes we see in enterprise and institutional website projects across Singapore and Southeast Asia. The framework, the two-phase delivery model and the gate between thinking and building are set out in full on the COMPASS methodology page.

Two ways to take the next step:

• Score your current site. The Website Scorecard is 21 questions and takes about six minutes. You get a score for each of the seven pillars, the order to fix them in, and a report you can put in front of leadership. No call required.
• Talk it through. If you are building the internal case for a redesign and want a view on sequencing and scope, start a COMPASS conversation.

Copy this checklist, adapt it, put it in your project plan. It is on this page in full for exactly that reason.