Redesigns fail when nobody can prove what changed. The site looks better, but the lead volume stays flat and the team starts guessing. That happens when you launch without a baseline.
A baseline does not need to be complex. It needs to show three things: how people find you, which pages create inquiries, and whether the site feels fast and reliable. If you capture that now, you can tell whether the redesign actually moved the business.
If you want a broader view of how a redesign pays for itself, pair this with the business website redesign ROI guide. That article explains the financial side. This one focuses on measurement.
Start with one clear definition of a lead
Most teams track pageviews and call it a day. That is noise. Decide what counts as a qualified inquiry first. For most service businesses, that is a completed contact form, a submitted project brief, or a booked call.
Google Analytics 4 lets you mark events as key events so they count as conversions. Google’s GA4 help doc explains how key events work and how you can mark them. See the GA4 key events guidance.
If you track leads as events, use a clear name like generate_lead. It is one of Google’s recommended events and keeps your reporting consistent. See the GA4 recommended event reference.
If you want help mapping this into your site structure, the contact page design guide shows where lead friction usually hides.
Capture visibility before you change content
A redesign almost always changes titles, headings, and internal links. That can shift rankings even if the content stays similar. Use Search Console now so you have a clean before‑and‑after view.
The Performance report shows clicks, impressions, CTR, and average position by page and query. Those are the baseline metrics that tell you whether visibility improved or dipped after launch. See the Search Console Performance report documentation.
Record your top pages and their top queries. If you lose traffic later, you will know which page changed and why it mattered.
Add light context to lead events
Knowing that a form was submitted is useful. Knowing which service or page triggered it is better. GA4 supports custom event parameters so you can pass details like form type, service interest, or page category. That makes the data usable when you ask which pages drive the best leads. See the GA4 event parameters guidance.
This is also how you connect site structure to lead quality. If a specific service page creates better leads, you can prioritize it in navigation and internal links.
Set a simple pre‑launch baseline
You do not need a dashboard jungle. A basic baseline sheet is enough. Capture:
- Total qualified inquiries per month
- The top three pages that drive those inquiries
- Search Console clicks and impressions for those pages
- A quick performance snapshot for each page
Measure the real pages on the devices and network conditions that matter to your audience. Keep the raw report or export with the date. An estimate is not a baseline.
Write the event contract before implementation
Tracking becomes hard to compare when names and parameters drift between pages. Write a small event contract that the old and new site can both follow. For each conversion event, define:
- the event name;
- the exact user action that triggers it;
- the page and form identifier;
- whether it is a key event;
- the fields that must never be sent;
- the analytics and product tools that should receive it.
For example, a form success should fire after the server accepts the submission, not when someone clicks the submit button. A booking event should represent a confirmed booking, not an opened scheduling widget. This distinction prevents a redesign from appearing successful because it generated more button clicks while creating no additional leads.
Keep business meaning separate from vendor syntax. You can map the same generate_lead outcome into GA4 and a product analytics event without inventing two definitions of a lead.
Separate discovery traffic from qualified intent
Pageviews, engaged sessions, and scroll depth help explain behavior, but they are not leads. Record them as supporting signals. Put higher-intent actions in a separate group:
- workflow or project intake started;
- contact form submitted;
- booking confirmed;
- pricing or process page reached from a relevant service page;
- return visit to a decision page.
Referral reporting needs the same discipline. Human visits from ChatGPT, Claude, Perplexity, or another assistant belong in an assistant-referral segment. AI crawler requests and automated agents do not. Server logs, crawler user agents, and analytics sessions answer different questions; combining them inflates the apparent audience.
The baseline should therefore show both acquisition and outcome: sessions by channel, qualified-intent events, and verified lead or booking events. If there are no verified conversions, record zero. Do not substitute a CTA click.
Preserve search evidence page by page
A sitewide traffic total is too coarse for migration decisions. Export Search Console page and query data for the period you will compare after launch. Keep:
- the canonical URL;
- clicks, impressions, CTR, and position;
- the strongest queries for that page;
- known redirect destination if the URL will change;
- important internal links and external links where known.
This tells you which pages can be removed safely, which should be consolidated, and which URL should remain stable even when the copy changes. It also gives you a diagnostic path after launch: a visibility drop can be traced to a page, query, redirect, or internal-link change instead of blamed on “the redesign.”
If content is being consolidated, annotate the baseline with the source-to-destination map. Compare the destination against the combined source set after search engines have processed the redirects.
Account for consent and data loss
Analytics is never a complete census. Consent choices, browser protection, ad blockers, form spam filtering, and cross-domain booking flows all change what is observable. Record the consent setup and domains included in the baseline so a configuration change is not mistaken for a behavior change.
Use first-party operational evidence to confirm high-value outcomes. A form delivery record or booking system confirmation can verify that a lead occurred; analytics explains the path around it. Keep personal data out of event payloads and use stable non-sensitive identifiers only where the privacy basis and retention rules are clear.
Protect the baseline when you redesign
The biggest mistake is launching and then rewriting tracking. Keep your lead events the same before and after launch. Otherwise your comparison breaks. If you must change events, document it clearly and keep an overlap period where both old and new events run.
This is where a good partner earns their fee. A serious build includes measurement planning, not just UI delivery. If you are comparing proposals, the proposal evaluation guide explains what to look for.
Use the baseline to make decisions, not just reports
After launch, validate the implementation immediately, but wait for a comparable window before claiming an outcome. Compare the first four complete weeks against the baseline and note seasonality, campaigns, or outages. If leads go up but search visibility drops, inspect changed content and internal links. If traffic stays stable but leads drop, inspect the conversion path and event implementation.
Use a short operating rhythm:
- Launch day: confirm consent, pageviews, source attribution, form success, and booking confirmation.
- Week one: inspect errors, broken paths, unexpected referral changes, and search coverage.
- Week four: compare the full baseline and investigate material changes page by page.
- Month three: decide what to keep, reverse, or test next.
Define thresholds before looking at the result. A broken lead event or lost high-intent landing page is an immediate fix. A small week-to-week traffic swing is a watch item, not a reason to rewrite the site again.
Keep the baseline in a small register with the metric definition, source report, date range, filters, owner, and known limitations. Save links or exports for the exact GA4, Search Console, product analytics, booking, and form views used. A number without its filters is not reproducible evidence.
When the definition changes, add a new row instead of silently overwriting the old one. This makes the comparison honest and gives the next operator enough context to continue the measurement rhythm.
Keep failed checks in the register too; missing or zero data is operational evidence, not an invitation to invent a positive result.
If you do not have internal ownership of this, keep it simple: check the baseline monthly and make one change at a time. The services overview and business websites service page give context for how a site should support lead quality, not just clicks.
If you want help setting the baseline or fixing tracking, you can start with contact or send a structured project brief. A redesign is expensive. A clear baseline is what makes it accountable.
