What Is a Conversion Funnel? How to Build One for Your Website

Your website may attract thousands of visitors a month while enquiries and orders stay flat. When a client shows me that picture, my first question is always the same: at which step do people actually leave? A conversion funnel is the most practical way to answer it. In this guide I walk you through the stages, how to map them to your own site and how to build the funnel exploration report in GA4.

What is a conversion funnel?

A conversion funnel is an ordered model of the steps a visitor takes on your website, from the first page view to a goal such as a purchase, a form submission or a booked call. Some people drop off at every step, and the funnel shows you where and how many.

In practice, the funnel metaphor rests on a simple observation. Many people enter at the top, and then fewer remain at each level below. For example, a hundred people view a product page. Some of them add it to the cart, fewer start checkout, and only a handful complete the order.

For me, the real value of a funnel is that it breaks one conversion rate into parts. A 1% conversion rate only tells you the outcome. The funnel, on the other hand, tells you whether the problem sits in ad quality, on the product page or in the checkout form. As a result, you spend budget and design effort on the step that needs it.

What are the stages of a conversion funnel?

Marketing literature offers several models. You may know AIDA (attention, interest, desire, action), the TOFU, MOFU and BOFU split, or awareness, consideration, decision and loyalty. The labels differ; however, the logic stays the same. In website projects I work with four stages, because each one maps to a real page and a measurable event.

The event names in the table are standard GA4 names. However, not every site gives equal weight to all four stages. On a single page campaign site, for instance, awareness and decision happen on the same screen. So instead of memorising a model, fit it to the way people really move through your site.

Why does the funnel model still work, and where does it fall short?

Admittedly, some marketers say the funnel is dead. Their point is fair: real customer journeys are not straight lines. People see your brand on Instagram, search for you on Google three days later and buy a week after that from an email. Others even leave the product page and message you on WhatsApp.

Still, the model works, because its job is not to draw the whole journey. Its job is to make the critical transitions on your site measurable. The rate from product page to cart, from cart to checkout and from checkout to order is concrete data you can track.

Knowing the limits keeps you from drawing the wrong conclusions. I keep three in mind:

• A funnel shows behaviour on your site, not channel attribution; that lives in a different report.
• Offline conversions such as phone calls, WhatsApp chats or in store sales do not enter the funnel on their own.
• On low traffic sites, percentages swing a lot; one bad week may not signal a real problem.

In short, use the funnel as a diagnostic tool rather than a full map of the customer journey.

I often explain this with a shop analogy. The funnel shows the aisle from the front door to the till. It does not show which ad the customer saw before walking in. For that second question you open attribution reports; for the first, you open the funnel. As long as you keep the two apart, the funnel gives you a reliable map and moves team debates onto data.

How do funnels differ for ecommerce, service and B2B websites?

Copying one funnel template to every website is one of the most common mistakes I see. The length, value and channel of the buying decision vary a great deal between industries.

In ecommerce, the funnel is at its clearest. Category page, product page, cart, checkout details and order confirmation follow one another. The steps are short, and the data is also plentiful. Therefore you can usually read the effect of a small improvement within a few weeks.

A service website has a shorter funnel, but the last step leaves the site. The visitor reads a service page, checks references and then fills in a form or calls. Specifically, the actual sale happens on the phone or in a meeting. In other words, the website funnel ends at the enquiry and the sales pipeline continues in your CRM.

In B2B, several people share the decision and it can take weeks or even months. Here I strongly recommend adding micro steps such as a quote request, a brochure download or a demo booking. Otherwise the long sales cycle makes the funnel look empty, and you end up fixing the wrong thing.

Which goal should your conversion funnel end with?

The last step of the funnel should be the action closest to the point where your business actually earns money. In ecommerce that usually means a completed order. For a service business it might be a qualified enquiry, a click to call or a confirmed appointment.

Let me clear up a frequent mix up. Newsletter sign ups, brochure downloads and video plays have value, but they are not the end of the funnel. Instead, treat them as intermediate steps. Otherwise a funnel that counts newsletter sign ups as conversions will look like a success while revenue stays flat.

You can also run more than one funnel on the same site. For example, a dental clinic site benefits from separate funnels for the booking form and for WhatsApp messages. Each funnel should lead to one goal; mixing two goals in one funnel makes the report unreadable.

Once you pick the goal, mark it as a key event in GA4. According to Google's documentation on key events, a newly marked event can take up to 24 hours to appear in standard reports. So do not look for numbers straight after setup.

How do you map funnel steps to the pages on your site?

A funnel that looks good on paper returns an empty report if it does not match your real page structure. Because of this, I start every project with a simple mapping table. The left column holds the funnel step, the middle column the URL or page group, and the right column the event that represents the step.

While you map, ask these questions in order:

1. Which pages do most visitors enter through? These form the top of your funnel.
2. Which pages sit on the shortest path to the goal?
3. Which button or form does the visitor use at each transition?
4. Does that click or form already fire an event in GA4?
5. Do you need a new event for any step that has none?

On a service site, for example, the map might read: blog post, then service page, then contact page, then form submission. Yet when you open GA4 you may find that most visitors go straight from the blog to the form. That is the moment to redraw the funnel around real behaviour rather than your assumption.

If mapping reveals gaps in your page structure, the funnel project often turns into a web design decision. You cannot put a service in a funnel if it has no page of its own.

Which events should you track for a conversion funnel?

In practice, a funnel report can only show the data you collect. Put simply, a step you do not measure will never appear in the report. As a result, your measurement plan becomes the skeleton of the funnel.

The GA4 list of recommended events gives you a solid starting point. I suggest you stick to the standard names, because some reports and the Google Ads integration recognise them. The main ones are:

• For ecommerce: view_item_list, view_item, add_to_cart, view_cart, begin_checkout, add_shipping_info, add_payment_info and purchase.
• For lead generation: generate_lead, qualify_lead, working_lead and close_convert_lead.
• For both: page views and form interactions from automatic or enhanced measurement.

On lead generation sites, firing generate_lead on form submission is not enough. When your sales team qualifies a lead, you can also send qualify_lead, and close_convert_lead when it turns into a customer. Then the funnel shows not only "form received" but also "became a real customer".

Tag your traffic sources properly while you are at it. If you label campaign links consistently with a UTM builder, splitting funnel losses by channel becomes much easier.

How do you build a funnel exploration report in GA4?

Funnel exploration is where most conversion funnel analysis in GA4 takes place. You open it from the Explore section. The setup runs like this:

1. Go to Explore in the left menu and pick the Funnel exploration template from the gallery.
2. Click the pencil icon next to Steps in the settings panel.
3. Give each step a clear name and define its condition, for example "event_name exactly matches add_to_cart".
4. Narrow a step with a parameter such as page path if you need to.
5. Order the steps, click Apply and choose the date range.
6. Add a breakdown dimension and segments to make the report comparative.

Google's funnel exploration help page says you can add up to 10 steps. My advice is to stay between four and six. Every extra step splits your users further, and on low traffic sites the last steps shrink to meaningless numbers.

Also keep one detail in mind. Each user enters the funnel only once, and the report counts only their first sequence of steps. So a person who adds to cart three times and buys once still appears as one user.

Should you use an open or a closed funnel?

First, know that GA4 gives you two options. In a closed funnel, users must start at the first step. In an open funnel, they can enter at any step.

So which one should you choose? The answer depends on your question. If you want to know how healthy the journey from product page to purchase is, use a closed funnel. If instead you want to see whether people enter the site at different points, such as landing straight on the cart from an ad, an open funnel tells you more.

My routine is simple. First I build a closed funnel to get baseline transition rates. Next I duplicate it as an open funnel with the same steps and compare the two side by side. A large gap tells me that many visitors join the funnel halfway through. That usually means it is time to rethink the landing page strategy.

Moreover, the gap between the two reveals where your ads send traffic. For example, a campaign link that points to the cart instead of the product page stands out at once in the open version.

What changes between "directly followed by" and "indirectly followed by"?

For each step in a funnel exploration you choose one of two connection types. With "directly followed by", the next step must happen immediately after the previous one. With "indirectly followed by", other actions can happen in between.

For most sites, however, the indirect option is more realistic. A visitor may look at another product, open a size guide or read reviews before adding to cart. Under the direct setting, all of those people look like drop offs, and your losses appear far larger than they are.

Instead, keep the direct option for narrow questions. For example, it makes sense for "do people who enter payment details complete the order right away?" You can also add a "within" time condition between steps. That lets you isolate, say, people who reach checkout within 30 minutes of adding to cart. I set that window based on the site's sales cycle, not as a fixed rule.

Finally, switch on "show elapsed time". It displays the average time between steps and shows where decisions slow down, which matters most in B2B funnels.

How do you read a conversion funnel report?

When the report opens, look at two numbers for each step. One is the number of users who reached it. The other is the share who left before the next step, which is the abandonment rate. The real story sits in the transition rates between steps.

Here is a worked example. The numbers are purely illustrative and do not belong to any client:

• Viewed a product page: 10,000 users.
• Added to cart: 800 users, an 8% transition rate.
• Started checkout: 400 users, which is 50% of those who added to cart.
• Completed the order: 200 users, which is 50% of checkout starters.

The overall conversion rate here is 2%. But the key question is which step would bring more orders if improved. Lift product to cart from 8% to 10% and, with the other rates unchanged, orders rise from 200 to 250. Improving the checkout step by the same relative amount has a similar effect. A small change in one rate can therefore make a big difference in orders. You can run these sums quickly with a percentage calculator.

That said, the cost of improvement differs from step to step. Weigh each step's potential gain against how hard it is to fix.

How do you find which step is leaking?

However, finding a step with high abandonment is not the end of the job. Now you move on to "why". GA4 offers three tools for this.

First, use a breakdown dimension. Split the report by device category and the gap between mobile and desktop jumps out. If checkout completion is 60% on desktop and 30% on mobile, the problem most likely lives in the mobile checkout form. Traffic source, country and landing page are also useful breakdowns.

Second, turn on "next action". It lists the top five actions users take after each step. If many people who leave the cart head to the shipping costs page, you can see the exact question on their mind.

Third, compare segments. According to Google's documentation you can compare up to four segments at once. I usually split new versus returning visitors and paid versus organic traffic.

In addition, I recommend confirming what the numbers suggest with session recordings. The numbers tell you where to look; the recordings show you what is actually happening.

How can you reduce drop off at each stage?

Once you find the leak, the fix depends on the stage where it happens. The list below sums up the interventions that most often work in practice. They are not guarantees, only starting points worth testing:

• Loss at awareness: the traffic does not match search intent. Review your target terms; the guide on finding keywords that drive sales helps here.
• Loss at consideration: the product or service page leaves questions open. Add pricing cues, delivery times, references and an FAQ.
• Loss at decision: the form is long, the cart holds surprises or trust signals are weak. Cut required fields, show shipping costs early and make payment options clear.
• Loss after action: the thank you page is empty. Spell out the next step, the expected response time and the contact channel.

Page speed affects every stage as well. A slow product page on mobile loses people before they even start considering. For that reason it makes sense to pair funnel work with a technical check. Within my SEO consulting work I often handle both on the same roadmap.

How do you widen the top of the funnel?

After you fix the leaks lower down, turn to the number of people entering the funnel. However, the goal is not simply more visitors; it is visitors with the right intent. A funnel widened with the wrong audience looks crowded at the top and produces the same sales at the bottom.

My order of priorities is this. First I find the landing pages that already convert well. Then I strengthen the channels that feed them: similar intent keywords in organic search and matching intent campaigns in paid search. Finally I test new channels with small budgets and track them as a separate funnel segment.

To judge quality at the top, break the funnel down by traffic source. If a channel brings many visitors and loses almost all of them at step one, its targeting or message does not match the landing page. In that case, fix the fit between message and page before you raise the budget.

In short, funnel width and funnel health are separate issues. Fix health first and widen second; the reverse usually wastes budget.

How can remarketing win back people who leave the funnel?

That said, people who drop out at any step are not lost for good. They know your site, and they have also shown interest. Therefore, remarketing exists to bring exactly this audience back.

Here your conversion funnel acts as a targeting map. You show one message to people who viewed a product but did not add it to the cart, and another to people who abandoned checkout. For instance, the first group may respond to an ad about product benefits. The second group is more likely to respond to a reminder about shipping or returns.

I cover audience setup and message sequencing in a separate article on strengthening your sales funnel with remarketing. When you track the return on that spend, a ROAS calculator speeds things up.

One warning, though: remarketing does not fix a structural problem in the funnel. If the checkout form is broken, the person you bring back gets stuck on the same form again. Close the leak first, then call people back.

How do you connect the website funnel with CRM data?

For service and B2B businesses, the GA4 funnel stops at form submission. What really matters to you, however, is how many of those forms turn into customers. You close that gap by joining the website funnel to the CRM pipeline.

The simplest method is to attach source data to every enquiry. Write UTM parameters and the landing page into hidden form fields. Then, when the lead lands in your CRM, its origin comes with it. In my own projects I log every enquiry with this data and follow the quote and closing stages there.

As a next step, you can send status changes from the CRM back to GA4. When a lead is qualified you send qualify_lead, and when it closes you send close_convert_lead, using the Measurement Protocol or server side tagging. Your funnel report then shows won customers next to form counts.

In addition, this connection exposes channels that bring many leads and few sales. If a campaign fills your inbox but almost none of those leads buy, the problem is targeting, not the on site funnel.

Which mistakes distort conversion funnel analysis?

Most funnel mistakes I have seen over the years are about method rather than technology. These come up most often:

1. Firing an event twice: a purchase event that fires again on page refresh inflates the end of the funnel.
2. Not filtering test orders and internal traffic: your team's trial runs look like real customers.
3. Looking at too short a date range: a few days of low volume data is mostly noise.
4. Mixing channels: paid and organic visitors arrive with very different intent, and a single funnel averages them.
5. Not logging changes: if you do not know when you changed the design or the price, you cannot explain a break in the report.

Above all, pay attention to the first point. A setup that sends the purchase event every time someone opens the thank you page makes the final step look higher than it really is. Consequently every rate you base decisions on is wrong. Place a test order after setup and check in DebugView that the event arrives only once.

How do you test and track funnel improvements?

Reading the result of a change correctly matters as much as making the change. Otherwise seasonality, a campaign period or a jump in ad spend can mislead you.

My tracking routine works like this. First I record the transition rate from the period before the change as a baseline. Next I ship the change on its own and change nothing else that week. After that I compare against a period of the same length and focus only on the rate of the step I changed.

If you have enough traffic, an A/B test is the most reliable method. With low traffic, a before and after comparison can work too. Still, treat that result as a strong signal rather than proof.

In practice, the trended funnel view in GA4 helps a lot here. When you switch from the standard view to the trended view, you follow each step's rate over time as a line chart. That way you can see clearly whether the rate breaks on the day the change went live. I also log ad changes on the same timeline and keep my Google Ads management reports next to the funnel report.

Where should you start with your conversion funnel?

After all this detail, let me reduce the work to a simple order. First, pick one business goal. Map the pages that lead to it and set up any missing events. Then build a closed funnel of four to six steps in GA4 and let at least a few weeks of data build up.

Once the data is there, find the step with the highest abandonment. Narrow down the cause with device and source breakdowns, make one change and watch the result in the trended view. When you repeat this loop every month, your funnel turns into an improvement list based on data rather than guesswork.

On ecommerce sites the loop tends to focus on cart and checkout. If you want to look at product, shipping and payment setup together, see my ecommerce consulting page. To feed the top of the funnel from social channels, the guide on driving website traffic from social media is a good companion.

If you would like to map your own funnel together, write to me through the contact page. In the first call we look at your current GA4 setup, see which steps go unmeasured and agree on which leak to fix first.