# Conversion funnel: turn website visits into customers

Map your conversion funnel from visit to customer. Use a worked example, a free worksheet and GA4 checks to find lost enquiries and choose the next fix.

Canonical page: https://cpcinsider.com/insights/ga4-lead-generation-funnel/
Author: Vladlens Kecko
Published: 2026-10-05
Updated: 2026-10-13


A conversion funnel shows whether people complete a defined sequence, such as viewing a service, starting an enquiry and sending it successfully. It helps identify where to investigate. It does not, on its own, explain why someone left or prove that a redesign will increase sales.

At CPCInsider, former Google specialist [Arturs Rubins](https://cpcinsider.com/about/#arturs-rubins) brings advertising-platform experience to a team connecting acquisition, website behaviour and CRM outcomes. For this task, the useful question is simple: which part of the journey deserves attention before you buy more traffic?

## Conversion funnel stages: connect the website to the business

A useful business funnel continues beyond the form: **relevant visit → valid enquiry → sales-accepted opportunity → new customer**. Each stage needs a definition, an owner and a record. “Valid” could mean a real, contactable request for your service; sales acceptance adds your own fit criteria. Agree those criteria before comparing campaigns.

Keep the website sequence in GA4 and the commercial stages in your CRM. They will not necessarily have identical counts: consent, identities, repeated visits and attribution differ. Do not divide CRM customers by GA4 users and present the result as a precise single-user journey without a validated join.

## Use this funnel diagnostic worksheet

This **hypothetical CRM-linked acquisition cohort** contains 800 attributable visits, 40 valid enquiries, 20 accepted opportunities and 10 customers. It is not a benchmark or a GA4 report.

| Transition | Example conversion | What to check | First practical action |
| --- | --- | --- | --- |
| Visit → valid enquiry | 40 / 800 = 5% | Intent, offer, proof and form delivery | Test the mobile journey and confirm the ad matches the page |
| Enquiry → accepted opportunity | 20 / 40 = 50% | Wrong service, location or qualification | Review rejection reasons and clarify eligibility |
| Opportunity → customer | 10 / 20 = 50% | Response, proposal and buying objections | Assign the next step and review lost reasons with sales |
| Visit → customer | 10 / 800 = 1.25% | All stages under the same cohort definition | Review the weakest supported hypothesis, then change one thing |

[Download the funnel worksheet](https://cpcinsider.com/resources/en/conversion-funnel-worksheet.csv). Enter your definitions, counts, owner, hypothesis and next review date. If the stages cannot be joined reliably, keep separate views and label the gap.


## 1. Choose a journey that reflects the website

Start with three steps: service-page visit, form interaction and confirmed submission. If customers can submit from several pages, define that explicitly. Do not require a homepage visit merely because it is the first page in your navigation. Do not mix a contact-form journey with an ecommerce checkout in the same funnel.

Use the [measurement-plan worksheet](https://cpcinsider.com/insights/advertising-measurement-plan/) to agree what each event means. Inspect a real successful test before using the events in a report. A missing event can look exactly like visitor abandonment.

## 2. Configure a focused exploration

In GA4, open Explore and choose Funnel exploration. Add named steps using the relevant event or page conditions. Choose a period with useful data and record the settings so another person can reproduce the report. Google's [funnel documentation](https://support.google.com/analytics/answer/9327974?hl=en) describes the configuration.

A closed funnel requires entry at the first step; an open funnel allows entry later. Choose based on the question, not whichever version looks better. Allow intervening actions when visitors legitimately read another page between steps. An unnecessarily strict sequence can exclude valid journeys.

## 3. Use drop-off as a diagnostic queue

The fictional example below uses people completing the same defined sequence. It is not a Latvian benchmark.

| Stage | People | What to investigate |
| --- | --- | --- |
| Service page | 200 | Does the ad promise match the page? |
| Form started | 40 | Is the offer and next step clear? |
| Enquiry confirmed | 20 | Are fields, errors or mobile interactions blocking completion? |

Here, 20 of 40 starters complete the enquiry: 50%. That calculation does not tell you whether the other 20 changed their minds or experienced an error. Inspect the form and your event implementation before choosing a remedy. Avoid comparing raw event counts with people in a sequence as if they were the same denominator.

## 4. Segment only when the sample supports it

Check mobile versus desktop and the language of the landing page. For a Latvia-focused site, poor performance on one translated page may reflect a different offer or broken form label rather than weaker customer demand. Keep acquisition-source definitions consistent when comparing traffic groups.

When a segment has only a few visitors, extend the observation window instead of declaring a winner. Pair the numbers with a manual walkthrough: can someone understand the offer, see the next step and correct an error without losing their input?

## 5. Connect the last step to lead quality

Twenty submissions are not twenty customers. Review accepted enquiries and won outcomes in the CRM separately. If more people finish the form but sales accepts fewer of them, the change may have reduced qualification too far. Keep website completion and commercial quality visible together.

Use the [landing-page optimisation checklist](https://cpcinsider.com/insights/landing-page-optimization-paid-ads/) to choose one focused improvement. Record the change date, compare similar conditions and allow for the sales cycle. A before-and-after comparison is a useful signal, not automatic proof of causality.

## Validate the events before blaming the page

Google's [recommended events](https://support.google.com/analytics/answer/9267735?hl=en) include `generate_lead` for submitted enquiries. It requires implementation; naming an event does not send it automatically. Use a confirmed successful submission rather than a button click as your success condition.

Enable debug mode on your test device and use [DebugView](https://support.google.com/analytics/answer/7201382?hl=en) to inspect the sequence. Test a successful submission, a validation error and a failed request. Check consent settings when events are absent; do not bypass a visitor's choice to make a report look complete.

### Is a sales funnel the same as a conversion funnel?

They overlap, but their boundaries differ. A website conversion funnel ends at a chosen action. A sales funnel usually follows enquiry handling and deals. Name the endpoint explicitly; a completed form is not a won customer.

### Which stage should we improve first?

Fix broken delivery or measurement immediately. Otherwise choose a stage with enough evidence and a plausible fix, not simply the largest percentage drop. A large drop can reflect intentional qualification. Use [CPC vs CAC](https://cpcinsider.com/insights/cpc-cpa-cac-differences/) to check costs and your [marketing plan](https://cpcinsider.com/insights/90-day-marketing-plan/) to assign the experiment.

## Find the next priority with CPCInsider

[Book a free 30-minute founder-led growth review](https://cpcinsider.com/?audit_from=%2Finsights%2Fga4-lead-generation-funnel%2F#audit). Bring your current plan, funnel or cost report. We will discuss the main growth constraint and follow up with a short written three-priority action plan. There is no obligation to hire CPCInsider; implementation is agreed separately.

Explore [tracking, attribution and CRM support](https://cpcinsider.com/services/tracking-attribution-crm/) for help connecting strategy, advertising and customer outcomes.
