Which Google Analytics Metrics Should You Choose to Grow Website Leads?

Which index to choose among the Google Analytics "data map" to grow leads through the website?

In marketing consulting, one of the most common mistakes is not lack of data, but too much data but not knowing which data is trustworthy, worth looking at, and worth acting on.

Google Analytics, Google Ads, Search Console, CRM, heatmap, call tracking, email marketing platform, landing page builder... each tool creates its own "universe of metrics". Without selective thinking, marketers can easily turn themselves into "report readers" instead of "growth decision makers".

My opinion

I did not choose the metric because it is available in Google Analytics. I chose the metric because it answers a specific business question.

With website lead generation, the biggest question is not:

How many visits does the website have?

Which is:

How many valuable leads is the website generating, from what source, at what cost, and where in the funnel is it losing customers?

That's the difference between data reporting and growth consulting.

Google Analytics is very powerful in telling us what users have done on the website. But if you only look at surface indicators such as users, sessions, page views, bounce rate, average session duration, marketers can easily fall into the "illusion of growth". Traffic may increase, but leads will not increase. Leads may increase, but leads are of poor quality. The form could be submitted more, but the sales team could not close any customers.

Therefore, I think the website measurement system should not start with the question "What indicators does GA have?", but must start with the question:

What business decisions need to be made from this data?

1. Not every metric that can be measured is worth tracking

A good dashboard is not one with many charts. A good dashboard is one that helps managers know:

• Should this channel's budget be increased?

• Should I cut the other channel?

• Which landing page is causing lead loss?

• Which traffic group has the highest conversion rate?

• Which lead should be prioritized for remarketing or transferred to sales?

• Is the current lead generation cost still reasonable compared to customer value?

The bottom line is: good metrics must lead to action.

If a metric only makes a report look "prettier" but doesn't change budget allocation decisions, optimize landing pages, adjust content or improve sales processes, then that metric is just decorative data.

2. I divided the index into 4 levels, instead of looking at them all equally

Level 1: Business performance index

This is the most important group of indicators, because it is directly tied to the ultimate goal of the website: generating leads and generating revenue.

To me, Qualified Leads is more important than Total Leads.

A campaign that generates 1,000 leads but only 20 quality leads is not necessarily better than a campaign that generates 200 leads but has 80 relevant leads. If businesses only optimize by the number of leads, it is easy for businesses to pour money into channels that create "junk leads".

Layer 2: Conversion funnel metrics

This group helps answer the question: where are users falling before becoming leads?

A basic website lead funnel might include:

Traffic to website
→ View landing page
→ Click CTA
→ Start filling out form
→ Submit form
→ Lead is recorded by CRM
→ Lead is confirmed by sales
→ Lead becomes a sales opportunity

Indicators that should be tracked:

In GA4, important actions such as form submission, booking, phone click or demo request should be configured as key events to more clearly measure website success. Google also allows using key events in GA to create conversions in Google Ads, helping to reduce measurement discrepancies between advertising and analytics. (Google Help)

Level 3: Behavioral diagnostic index

This is a group of indicators that help explain why results are good or bad.

Example:

One point to note: in GA4, bounce rate is the opposite of engagement rate, meaning the session rate is not considered engaged. Therefore, you should not take the old benchmark like "bounce rate below 40% is good" and apply it mechanically to every website, every industry and every type of page. (Google Help)

For example, a blog post with a high bounce rate is not necessarily bad if the user reads it, understands the problem and then leaves. But a landing page running ads with high bounce rate, low CTA clicks and low start form is a very worrying signal.

Layer 4: Data quality indicators

This is a layer that many people ignore, but it is extremely important.

If the data is wrong, all subsequent analysis is wrong.

I often track additional indicators such as:

A professional measurement system doesn't just ask "lead increased or decreased", but also asks:

Is this lead data clean enough to trust?

3. Metrics I don't consider key KPIs

There are some metrics that are still useful, but I wouldn't include them as primary KPIs if the goal is lead growth.

3.1. Sessions

Sessions indicate how many visits the website has, but do not specify whether that traffic is valuable or not.

If sessions increase by 50% but leads do not increase, the business may be attracting the wrong audience. If sessions are down 20% but qualified leads are up, maybe your targeting strategy is getting better.

So sessions are just input, not results.

3.2. Page Views

High page views may be due to attractive content, but it may also be because users have to go around because they cannot find the necessary information.

I only use page views when doing in-depth analysis according to:

• content group;

• landing page;

• pre-conversion journey;

• content supports lead generation;

• conversion rate from article to service page or form.

3.3. Average Session Duration

Average session time can be useful, but should not be interpreted in absolute terms. Users stay for a long time maybe because they are interested, but also because the content is confusing, the form is confusing, or the journey is unclear.

This indicator should go along with CTA click, scroll depth, form start and conversion rate.

3.4. Bounce Rate

Bounce rate should not be deified. With lead generation websites, I prioritize looking at engagement rate by traffic source and landing page rather than looking at the overall bounce rate of the entire website.

A website with many market education blog posts will behave differently than a direct sales landing page. If you add them all up, the average will obscure important insights.

4. After choosing the index, how should the data be processed to be usable?

Choosing the right index is just the first step. Data is only truly valuable when processed into a system that can be analyzed, compared and acted upon.

I usually process in 7 steps.

Step 1: Standardize measurement goals

Before tagging, it is necessary to clearly define:

• What is Lead?

• Which lead is considered a valid lead?

• Which action is a macro conversion?

• Which action is only a micro conversion?

• Is the lead recorded in GA, CRM or both?

• When is a lead considered MQL?

• When is a lead considered SQL?

• Which leads were eliminated because of spam, wrong needs or not qualified?

Example:

Macro conversion:
- Submit consultation form
- Schedule demo
- Call hotline from mobile
- Message WhatsApp/Zalo
- Register for trial
Micro conversion:
- Click CTA
- Scroll 75%
- View price list
- Download brochure
- View case study
- View contact page

If not clearly defined, each department will understand "lead" in a different way. Marketing reports a lot of leads, sales reports poor leads, management doesn't know who to trust.

Step 2: Design structured event tracking

In GA4, data is organized around events. Therefore, event tracking must be designed as a common language between marketing, analytics and sales.

Some events should include:

page_view
cta_click
form_start
form_field_error
form_submit
phone_click
zalo_click
whatsapp_click
file_download
pricing_view
case_study_view
demo_request
lead_created

The important point is not only tracking event names, but also tracking event parameters.

For example with cta_click, there should be:

cta_text
cta_position
page_location
page_type
campaign_name
device_category

With form_submit, there should be:

form_id
form_name
landing_page
lead_type
service_interest
traffic_source
campaign_name

However, special care should be taken not to send personally identifiable information such as email, phone number, or full name directly into Google Analytics. Google has a Personally Identifiable Information policy, and businesses should handle identification using internal IDs or pseudonymous codes instead of pushing personal data into GA. (Google Help)

Step 3: Normalize UTM and traffic source

One of the biggest reasons GA data is difficult to use is arbitrarily placed UTMs.

For example, the same Facebook channel can appear in many forms:

facebook / cpc
Facebook / paid
fb / ads
meta / paid_social
m.facebook.com / referral

At that time, the power source was reported broken. Managers do not know how many leads Facebook actually generates, how much CPA, how much conversion rate.

I usually standardize UTM according to the structure:

utm_source = platform
utm_medium = channel type
utm_campaign = campaign
utm_content = content variation
utm_term = keyword or audience

Example:

utm_source=facebook
utm_medium=paid_social
utm_campaign=leadgen_q3_2026
utm_content=video_testimonial_a
utm_term=lookalike_3percent

The rule is: UTM must be written as data for analysis, not for show.

Step 4: Data cleaning

Raw data from GA and measurement tools should not be used to draw conclusions.

Need to clean up common errors:

• internal traffic type;

• type of bot or referral spam;

• check duplicate form submission;

• check event fired multiple times;

• check if thank-you page reload creates fake leads;

• check cross-domain tracking;

• check referrals from payment gateway, CRM, landing page tool;

• check campaign for missing UTM;

• check the difference between GA and CRM.

A very real example: if the form submit event is triggered every time a user reloads the thank-you page, the report will show an increase in leads, but in reality the business has no more leads.

Wrong data is usually not loudly wrong. It goes wrong silently, and causes businesses to make wrong decisions for many months.

Step 5: Connect GA to CRM

This is the step that turns analytics from "measuring website behavior" into "measuring business performance".

GA said:

• where the user comes from;

• which page they look at;

• what do they click;

• which form they submitted;

• what equipment do they use;

• which campaign they came from.

CRM says:

• Is that lead valid;

• Can sales contact me;

• Does lead have the right customer profile;

• will the lead become an MQL;

• will lead become SQL;

• Does the lead generate revenue?

If these two systems are not connected, marketing can easily be optimized by lead volume instead of lead quality.

An ideal data table should have the form:

date
user_pseudo_id / client_id
session_id
source
medium
campaign
landing_page
event_name
form_name
lead_id
lead_status
mql_status
sql_status
deal_value
revenue

GA4 can export data to BigQuery for advanced analysis, and Google also describes combining GA data with other sources like CRM in BigQuery for reporting and deeper analysis. (Google Help)

Step 6: Convert data into analytical model

Once the data is clean and connected to CRM, it is time to run analytical models.

6.1. Funnel Analysis

The goal is to find the drop point in the lead journey.

Example:

1. Landing page views: 10,000
2. CTA clicks: 1,800
3. Form start: 900
4. Form submission: 300
5. Qualified leads: 120
6. Sales opportunity: 40

From here, it is possible to detect:

• CTA is not attractive enough if landing views are high but CTA clicks are low;

• form is too long if form start is high but submit is low;

• targeting is wrong if form submission is high but qualified leads are low;

• follow-up sales are poor if qualified leads are high but opportunities are low.

GA4 has Funnel Exploration to analyze the steps users go through and where they fail or drop out of the funnel. (Google for Developers)

6.2. Path Analysis

Path analysis helps see which journey users actually follow, instead of assuming they will follow the exact journey the marketer draws.

For example, the marketer thinks the user will go:

Homepage → Service Page → Pricing Page → Contact Form

But the data may show that the actual journey is:

SEO Blog → Case Study → About Us → Pricing → Exit

Or:

Google Ads Landing Page → Form Start → Form Error → Exit

GA4 Path Exploration illustrates the event stream and the screens/pages the user viewed or actions performed, making it very useful for detecting actual behavior. (Google Help)

6.3. Attribution Analysis

Attribution helps answer the question: which channels contribute to leads?

But I don't consider attribution to be the "absolute truth". I see it as a perspective.

Should compare multiple models:

• first touch;

• last touch;

• linear;

• time decay;

• data-driven if there is enough data;

• CRM-based attribution.

For example, SEO may not be the last channel to close leads, but it is the first channel to create awareness. Remarketing can close leads well, but without previous organic content, users may not have enough trust to convert.

So the goal is not to find a perfect attribution model, but to understand the role each channel plays in the decision-making journey.

6.4. Lead Scoring

When the data is good enough, a lead scoring model can be built.

Example of a simple lead point:

Lead Score =
35% high-intent behavior
+ 25% traffic source quality
+ 25% ICP relevance
+ 15% content engagement depth

Signals may include:

• see pricing page;

• see case study;

• return to the website many times;

• comes from keywords with high intent;

• submit consultation form;

• choose high-value services;

• belongs to the target industry;

• the company has an appropriate size.

Lead scoring helps marketers not only ask "who left the information", but ask deeper:

Which lead is worth prioritizing sales first?

Step 7: Turn data into a decision-making dashboard

The final dashboard should not have too many indicators.

For lead generation websites, I usually recommend less than 10 main KPIs:

More importantly, the dashboard must have an operating rhythm.

Example:

Weekly:
- Which channels increase/decrease leads?
- Which landing pages decrease conversion rate?
- Which campaigns have abnormal CPL?
- Which forms have high abandonment?
Monthly:
- Which channels generate the best qualified leads?
- Which channels create the best pipeline?
- Should the budget be increased/decreased?
- Do we need to test a new landing page?
Quarterly:
- Is CAC still reasonable?
- Has CLV changed?
- Should you prioritize acquisition or retention?
- Do you need to change your content/channel strategy?

5. Data is only valuable when tied to decisions

After selecting and processing data, the final question is always:

What does this data cause us to do differently?

Example:

This is where analytics becomes growth.

Not because we have more charts, but because we know exactly where to intervene.

6. A note about A/B testing

A/B testing is still a very important tool in optimizing website leads. However, it is important to update that Google Optimize and Optimize 360 ​​have been discontinued since September 30, 2023. Therefore, when building a current testing system, businesses should use other A/B testing platforms or deploy testing through landing page platforms, CMS, product analytics or specialized testing tools. (Google Help)

But tools are not the core point. The core point is experimental thinking:

Hypothesis → Design variation → Run tests → Measure key events → Assess impact → Deploy or remove

Example:

Hypothesis:
Landing page has too much technical information, making users afraid to fill out the form.
Test:
A: Current Landing page
B: New landing page, focusing on pain points, case studies and CTA more clearly
Main metrics:
Form submission rate
Secondary metrics:
CTA click rate, form start rate, engagement rate, qualified lead rate

A good test doesn't just ask “which version has more leads”, but also asks:

Which version generates better quality leads?

7. Conclusion: Choose fewer metrics, but choose better

In Google Analytics and marketing measurement tools, data is a lot. But growth doesn't come from looking at more metrics. Growth comes from choosing the right metrics, handling them the right way, and turning them into concrete decisions.

My opinion can be summarized as follows:

Don't build a dashboard to prove that marketing is busy. Build a measurement system to prove marketing is creating value.

With website lead generation, I will not start with sessions, page views or bounce rate. I'll start with lead, qualified lead, conversion rate, cost per qualified lead, lead quality by source, funnel drop-off and pipeline value.

After that, I will clean the data, normalize events, normalize UTM, connect GA with CRM, analyze funnels, analyze journeys, build lead scoring and put it all into a decision-making dashboard.

Because in the end, marketing data is only truly meaningful when it helps businesses answer three questions:

1. Where should the money go?
2. Where is the bottleneck?
3. Which leads are worth pursuing?

When you answer those three questions, Google Analytics is no longer a traffic reporting tool.

It becomes a growth-oriented system.