Website Redesign Measurement Plan: 40 KPIs and a Before-and-After Scorecard

A website redesign can look better and still make the business worse.

That sounds harsh, but it happens. The new site wins compliments while qualified leads fall, organic landing pages disappear, forms stop recording attribution, or mobile visitors struggle with a polished animation nobody tested. Six months later, the team has screenshots and opinions but no defensible answer to a basic question: Did the redesign work?

This measurement plan fixes that problem. It gives business owners, web teams, and agencies a shared scorecard for the period before launch, the unstable weeks immediately after launch, and the months when a fair business comparison becomes possible.

The central rule is simple: agree on the evidence before anyone sees the new design.

The Website Redesign Scorecard at a Glance

Use one primary business outcome, three to five supporting outcomes, and a set of guardrail metrics that must not get worse. Forty measures are listed below, but no sane team should call all 40 KPIs. Most are diagnostic measures.

This hierarchy matters. A lower bounce rate isn’t a win if qualified leads fall. More form submissions aren’t a win if spam doubled. Faster pages don’t excuse lost revenue, but speed can explain why revenue changed.

The UK Government Service Manual recommends defining a small set of performance metrics, continuously measuring them, and setting future targets instead of relying on isolated snapshots (GOV.UK). That is a strong operating model for commercial websites too.

Step 1: Write the Redesign Hypothesis

Before choosing metrics, finish this sentence:

We believe changing [part of the website] for [specific audience] will improve [business outcome] because [evidence]. We will consider it successful when [target] improves without [guardrail] declining.

Example:

We believe simplifying service navigation for operations managers will increase qualified quote requests because customer interviews show that buyers cannot tell which service fits their job. We will consider it successful when sales-accepted quote requests per 1,000 sessions rise by 15%, without organic clicks to service pages falling by more than 5%.

That statement forces four useful decisions: who the redesign serves, which behavior should change, why the team expects it to change, and what damage is unacceptable.

Avoid goals such as “modernize the brand,” “improve engagement,” or “make the site easier.” They can be valid design intentions, but they aren’t measurable business outcomes. Translate them. “Make the site easier” might become “raise the task-completion rate for finding the correct replacement part from 58% to 75%.”

Step 2: Establish a Defensible Baseline

A baseline is not last month’s dashboard. It is a comparison period selected to reduce distortion.

Use at least eight weeks of pre-launch data for a steady lead-generation site. Use 12 months when the business is seasonal. Compare the same weekdays, exclude known outages and internal traffic, and annotate campaigns, price changes, product launches, staffing shortages, or tracking changes.

Record both totals and rates. A site can produce fewer leads because traffic fell, even while its conversion rate improved. Conversely, total leads can rise because ad spending increased while the site itself converts worse.

Capture these baseline fields for every chosen KPI:

Use a median when one unusual campaign or outage could skew the average. Keep the raw weekly values so nobody mistakes normal variation for redesign impact.

40 Website Redesign Metrics, Organized by Decision

Choose the few measures that match the redesign hypothesis. The rest help diagnose why the primary outcome moved.

Business and Sales Outcomes

1. Qualified leads: submissions or calls that meet agreed sales criteria.
2. Sales-accepted lead rate: qualified leads accepted by sales divided by leads delivered.
3. Pipeline created: potential contract value tied to website-sourced opportunities.
4. Closed revenue: revenue attributed to website-originated customers.
5. Revenue per session: online or attributed revenue divided by sessions.
6. Cost per qualified lead: relevant acquisition and site cost divided by qualified leads.
7. Booked appointments: completed bookings, not clicks on a scheduling button.
8. Quote-to-sale rate: won quotes divided by website-originated quotes.

Connect analytics to the CRM wherever possible. A thank-you-page visit is a useful event, but it cannot tell you whether the inquiry had budget, fit, or buying intent. The GOV.UK performance model similarly distinguishes completion, cost per transaction, digital take-up, and user satisfaction rather than treating one interaction as the whole outcome (GOV.UK).

Conversion and Journey Measures

9. Primary conversion rate: primary outcomes divided by eligible sessions or users.
10. Landing-page conversion rate: primary outcomes by first page viewed.
11. Form-start rate: form starts divided by visitors who saw the form.
12. Form-completion rate: successful submissions divided by form starts.
13. Field error rate: validation errors divided by form interactions.
14. Call conversion rate: qualified tracked calls divided by eligible visits.
15. CTA click-through rate: meaningful CTA clicks divided by views.
16. Journey completion rate: users reaching the intended final step divided by journey starters.
17. Internal search success: search sessions ending in a useful action without repeated reformulation.
18. Assisted conversion rate: conversions where a key content page appeared earlier in the journey.

Don’t use an overall conversion rate alone. Segment it by device, source, new versus returning visitors, and major landing-page family. A redesign that helps desktop paid traffic but harms mobile organic traffic can appear neutral in the blended number.

Search and Acquisition Measures

19. Organic clicks: clicks from search results, measured in Search Console.
20. Organic impressions: visibility before the click.
21. Nonbrand organic clicks: search demand not containing the company or product brand.
22. Top landing-page clicks: organic clicks to the pages that historically created demand.
23. Indexed priority URLs: important canonical pages present in Google’s index.
24. Valid referring domains retained: linking domains still resolving to relevant pages.
25. Paid landing-page conversion rate: conversions from paid sessions by destination.
26. Email and referral conversion rate: outcomes from owned and partner traffic.

Google says site moves with URL changes can cause temporary ranking fluctuation while its systems recrawl and reindex the new URLs (Google Search Central). That makes launch-week search performance a monitoring signal, not a final verdict. Google also recommends permanent server-side redirects such as 301 or 308 redirects for permanent moves (Google Search Central).

Track priority URLs individually. A stable sitewide click total can hide the loss of a high-intent service page while low-value blog traffic grows.

Experience, Performance, and Quality Measures

27. Task-success rate: participants who complete a defined task without help.
28. Time on task: time required for a successful attempt.
29. Single Ease Question score: post-task rating of perceived difficulty.
30. Largest Contentful Paint: loading performance at the 75th percentile.
31. Interaction to Next Paint: responsiveness at the 75th percentile.
32. Cumulative Layout Shift: visual stability at the 75th percentile.
33. JavaScript error rate: sessions or page views affected by client-side errors.
34. 404 response rate: missing-page requests divided by eligible requests.
35. Accessibility defects: confirmed issues by severity and template.
36. User satisfaction: consistent post-journey or post-support rating.

For Core Web Vitals, Google’s current “good” thresholds are LCP within 2.5 seconds, INP within 200 milliseconds, and CLS at or below 0.1, assessed at the 75th percentile (web.dev). Field data matters because lab tests cannot reproduce every visitor’s device, network, and behavior.

Performance belongs on the business scorecard, not only the developer report. Rakuten 24 reported a 33.13% increase in conversion rate and a 53.37% increase in revenue per visitor after its Core Web Vitals work, while the study also found that visitors with good LCP converted at a higher rate (web.dev). That does not promise the same lift for another site. It shows why speed should be tested against commercial outcomes.

Operational Guardrails

37. Analytics event coverage: tested required events divided by required events documented.
38. Consent-adjusted measurement rate: observable sessions or conversions after consent effects.
39. Uptime and failed-request rate: availability plus server errors during key journeys.
40. Content accuracy defects: incorrect prices, contacts, claims, locations, links, or legal text.

These measures catch quiet failures. A new form might submit correctly but omit the campaign source from the CRM. A phone number may work on desktop and be wrong in a mobile sticky bar. Revenue reporting can look weak simply because the redesign broke cross-domain tracking.

Step 3: Build a Three-Window Evaluation Plan

One report date cannot answer every question. Use three windows.

Window 1: Launch Control, Days 0 to 7

The job is damage control. Monitor uptime, server errors, 404s, robots directives, canonical tags, analytics events, forms, calls, checkout, page speed, and priority redirects. Compare hourly and daily patterns with normal ranges, but don’t declare a conversion winner.

Set explicit rollback thresholds before launch. Examples include checkout failure, loss of lead routing, accidental noindex directives, or a severe decline on the main revenue path. Assign one person the authority to call a rollback.

Window 2: Stabilization, Days 8 to 30

Now look for directional changes by segment. Review Search Console coverage and organic landing pages, CRM lead quality, form errors, mobile conversion, internal search terms, call recordings, and support questions.

Fix defects without bundling in a new creative redesign. If the team changes the offer, traffic mix, navigation, and form during the same week, the final analysis will be guesswork.

Window 3: Business Evaluation, Days 31 to 90+

Compare the agreed post-launch period with the baseline and a seasonally appropriate control. Report the primary outcome, supporting outcomes, guardrails, and material outside factors. Long sales cycles require pipeline reporting first and closed revenue later.

Quantitative data tells you what changed. Usability testing helps explain why. Nielsen Norman Group defines usability testing around representative participants performing representative tasks, with the researcher observing behavior and listening for feedback (Nielsen Norman Group). Test the old experience before redesign decisions when possible, then repeat the same tasks after launch.

How to Separate Redesign Impact From Everything Else

Perfect attribution is rare. Better evidence is still possible.

Use matched periods and segment results. If both desktop and mobile traffic changed similarly but only mobile conversion fell, investigate mobile experience. If conversion stayed steady within every channel but the overall rate changed, the cause may be a different channel mix. If redesigned templates improved while untouched templates stayed flat, that strengthens the case that design contributed.

When traffic permits, release the redesign to a portion of users or use an A/B test for high-risk components. When it doesn’t, keep a clear change log and annotate campaigns, pricing, inventory, sales staffing, weather events, news coverage, and tracking repairs.

Report uncertainty in plain language:

Qualified leads per 1,000 sessions increased 18% against the eight-week baseline. The increase was concentrated on redesigned service pages and appeared across organic and paid traffic. A concurrent 9% increase in branded search may have contributed, so we attribute the full lift to the redesign only with moderate confidence.

That is more credible than claiming every improvement as a design win.

The Copy-Ready Redesign Report

Use this structure for the 30-, 60-, and 90-day review:

1. Decision: Keep, optimize, investigate, or roll back.
2. Primary outcome: Baseline, target, current result, percentage change.
3. Supporting outcomes: Three to five measures with the same comparison.
4. Guardrails: Pass, watch, or fail, with the owner of each repair.
5. Segment findings: Device, channel, landing-page group, audience, geography.
6. Evidence quality: Tracking gaps, sample limits, seasonality, outside changes.
7. User evidence: Task-test findings, survey themes, sales and support feedback.
8. Next actions: Owner, deadline, expected effect, and measurement method.

The report should fit on one page before appendices. A useful scorecard leads to decisions. A 70-slide analytics tour usually delays them.

Common Questions

How long should you measure a website redesign?

Monitor technical guardrails immediately, but allow at least 30 to 90 days for a business evaluation on most lead-generation sites. Seasonal businesses should compare against the same period from the prior year. Sites with low traffic or long sales cycles need more time.

What is the best KPI for a website redesign?

The best KPI is the business outcome the redesign was meant to improve, such as sales-accepted leads, booked appointments, pipeline, or revenue per visitor. Conversion and experience metrics should explain that outcome, not replace it.

Should bounce rate be a redesign KPI?

Usually not as a primary KPI. Its meaning changes by page purpose, tracking configuration, and analytics platform. Use journey completion, qualified conversion, task success, and landing-page outcomes instead.

What if tracking was broken before launch?

Don’t manufacture a baseline. Use CRM records, call logs, Search Console, server logs, invoices, and pre-launch usability tests where available. Label the gap, repair measurement, and set the first reliable period as the new baseline.

Who owns redesign measurement?

One accountable owner should maintain definitions and publish the scorecard, but sales, marketing, development, and operations must validate the measures they create or consume. Measurement is part of the redesign scope, not a cleanup task after launch.

Make the Next Redesign Prove Its Value

A good redesign plan says what will change. A strong one says how the business will know it worked.

If you want a website built around qualified leads, measurable journeys, and the operational details that keep results intact after launch, start a conversation with Your Web Team.