How to measure whether a website redesign worked

The starting point

A website looks better after a redesign. The team likes the new pages. A few people compliment the motion. None of those observations proves the project helped the business.

The useful question is what changed in the journey from visitor to meaningful action.

Start measuring before launch, not after the first month.

Save a baseline before changing the site

Pick the business action the website exists to support. For a professional services company, the action might be a qualified enquiry. For SaaS, a demo request or trial signup. For a local service business, a phone call or booking.

Then save the current data for a defined period:

• Sessions by traffic source.
• Organic search clicks and impressions.
• Important landing pages.
• Contact or demo CTA clicks.
• Confirmed form submissions.
• Qualified leads, if tracked.
• Mobile versus desktop behaviour.
• Major technical issues.

Also note current campaigns, seasonality, pricing changes, product launches, or sales changes. Those factors affect results after launch. A redesign should not receive credit or blame for every movement in the business.

Track the action, not merely the page view

A page view tells you somebody opened a page. A CTA click tells you the next step attracted attention. A verified submission tells you someone completed a request. A qualified lead tells you the request fits the business.

Those are different measurements.

Google Analytics lists generate_lead as a recommended event for a submitted request and offers further events for stages such as qualify_lead. Use event names and setup appropriate to the property rather than assuming every stage appears automatically.

If the form sends data to a CRM, test the whole handoff. A submission showing in analytics without a real enquiry reaching the team creates misleading confidence.

Choose a small scoreboard

I would usually start with three measures rather than twenty.

• 1. Qualified enquiry volume or another main business action.
• 2. Conversion rate for the relevant landing pages or traffic segments.
• 3. A supporting signal tied to the redesign goal, such as organic clicks, form completion, or demo interest.

Add behavioural evidence from Microsoft Clarity when sessions exist. Heatmaps and recordings help locate confusing buttons, mobile friction, dead clicks, or sections buyers fail to reach.

Do not treat a recording as proof of what every visitor thinks. Use behaviour as evidence for a hypothesis, then test the fix.

Compare like with like

A new campaign starting the same week as the redesign changes traffic composition. A new location, pricing change, seasonality, or marketing push changes buyer behaviour.

Compare similar time windows when possible. Segment organic, paid, referral, and direct traffic. Review the same key actions on both sides of launch. Check whether tracking definitions remained consistent.

For a redesign involving URL changes, keep separate watch on organic landing pages and indexing. A technical migration issue deserves investigation even if the design looks finished.

Watch the weeks after launch

During the first week, confirm functional health. Are important URLs accessible? Do forms send correctly? Does Search Console show expected canonicals? Are tracking events firing?

After enough data accumulates, review early behaviour and query trends. A meaningful business comparison takes longer when the site receives only a small number of visits or leads.

Avoid changing the site every day because a dashboard moved by one event. Write down the question and wait for enough information to judge the answer.

Decide what deserves the next change

The review should end with a decision.

If qualified traffic is weak, the next project might be search or distribution rather than another design round.

If traffic is relevant but buyers do not act, review the offer explanation, proof, CTA, and mobile flow.

If leads arrive but sales follow-up is slow, the problem has moved outside the page.

A successful redesign is easier to defend when the work has measurable goals, honest attribution limits, and a useful next decision.