How to Read a Conversion Funnel and Fix the Biggest Leak

What a Funnel Actually Measures

A conversion funnel is a sequence of steps between a visitor and a payment. Each step has an entry count and an exit count. The difference is your leak. That is it. The complexity people add on top (multi-touch attribution, assisted conversions, micro-funnels) is useful eventually, but it is noise until you know where the main drain is.

A typical SaaS funnel looks like this: landing page → sign-up page → onboarding → first key action → paid conversion. An ecommerce funnel is usually: product page → add to cart → checkout → payment confirmation. The exact steps depend on your product, but the principle is the same: you need a number for each transition.

What you are looking for is the step with the worst drop-off rate relative to the revenue it is blocking. A 40% drop-off from pricing to sign-up is not the same as a 40% drop-off from sign-up to first use, because the first group was buyer-intent traffic and the second group might have been curious tire-kickers. Context matters. Dollar value matters more than percentages.

How to Build a Funnel Worth Reading

Start by writing your funnel on paper before you touch any analytics tool. List every step a user takes from first touch to paid customer. Include the steps that feel obvious, like the email confirmation click or the credit card entry screen. Those obvious steps are where surprisingly large leaks hide.

Then instrument each step as a distinct event or pageview. If your analytics tool cannot tell you how many people entered step three and how many exited to step four, you do not have a funnel yet; you have a list of pages. The transition rate between consecutive steps is the number that matters.

One thing I see constantly: founders measure traffic to the pricing page but not traffic from the pricing page to the sign-up page. So they know pricing gets 2,000 visits a month but they have no idea that 1,700 of those people leave without clicking anything. That 85% drop-off is the story. The 2,000 visits number is flattering noise.

Finding the Biggest Leak

The biggest leak is not necessarily the step with the highest drop-off percentage. It is the step where the combination of drop-off rate and traffic volume destroys the most revenue. You find it by multiplying.

1. List every funnel step with its entry count and exit count for the same time period.
2. Calculate the drop-off rate at each step: (entries minus exits) divided by entries.
3. Estimate the revenue blocked at each step: multiply the number of people who left by your average revenue per conversion.
4. Rank the steps by revenue blocked, not by drop-off rate.
5. The top item on that ranked list is your only job right now.

Here is a concrete example. Say your sign-up page converts at 60%, which is pretty good. Your pricing page converts at 20%, which sounds bad. But if 500 people hit the sign-up page and 3,000 hit the pricing page, the pricing page is losing 2,400 people versus 200 people on the sign-up page. At a $50 average sale, the pricing page leak is worth $120,000 in potential monthly revenue. The sign-up page leak is $10,000. You do the math.

Diagnosing Why People Leave

Once you know which step is the biggest leak, you need to understand why. There are three ways to do this and you should use all three before writing a single line of code.

Behavioral Data

Heatmaps and click maps show you where attention goes and where it does not. On a pricing page, do people scroll past the plan comparison? Are they clicking on a feature name that is not a link, which means they want to know more but there is no path to learn it? Are there rage clicks on an element that looks interactive but is not? These patterns are invisible in a pageview count and obvious in a click map.

Scroll depth is particularly underused. If 70% of visitors never see your call-to-action because it sits below the fold on mobile, that is a layout problem, not a messaging problem. Fixing the layout takes an afternoon. Rewriting the copy takes a week and might not move anything.

Exit Surveys and Sales Conversations

Put a one-question exit survey on the page that is leaking. Ask something specific: "What stopped you from signing up today?" Give four options that reflect your real hypotheses (pricing, missing feature, not sure if it fits my use case, just browsing) and a free-text field. Run it for two weeks. You will hear the same two or three answers repeatedly. Those are your real conversion blockers.

If you have any sales conversations happening, read the transcripts for the last twenty deals that did not close. The objections in those transcripts are the objections on your page that your copy is not handling. Fix the copy to handle the objection, not to sound better.

Where Traffic Comes From Matters

A 20% conversion rate from organic search and a 20% conversion rate from cold paid traffic are not the same problem. Organic visitors read your content before they arrive; paid visitors may have no idea what your product does. If you are segmenting your funnel by traffic source and you are not, you should be, because you might find that the leak only exists for one channel. That changes the fix entirely.

Fixing It Without Guessing

Now you have a ranked list of leaks, behavioral data on what people do on the leaking page, and qualitative data on why they leave. You can make one focused change instead of a spray of hopeful tweaks.

Do not run an A/B test unless you have enough traffic to reach statistical significance in under four weeks. If you have 200 visitors a month on a page, A/B testing will take six months to tell you something. Make the change, watch the metric for three to four weeks, and decide. The speed of learning matters more than experimental purity when you are small.

The fixes that move conversion the most, in my experience: removing friction from the sign-up flow (fewer fields, no credit card required), answering the most common objection above the fold, and making the value proposition specific rather than clever. "Cut your customer acquisition cost in half" beats "Grow smarter" every time.

What to Look For in Your Analytics Tool

You need three things from your analytics: funnel visualization with per-step drop-off rates, behavioral data (heatmaps, click maps, scroll depth) on the leaking pages, and revenue attribution so you know which traffic sources produce buyers versus browsers.

Most tools give you the funnel visualization. Fewer give you revenue attribution that actually connects pageviews to payments. That gap is why Conclick exists: I was stitching together three separate tools to answer one question, which traffic source, campaign, or funnel step is producing paid customers. The revenue attribution in Conclick connects directly to Stripe, Paddle, Polar, Lemon Squeezy, and Dodo, so every payment traces back to the source and funnel step that earned it. The heatmaps and click maps are real-screenshot overlays, not wireframe approximations, and the auto-detected funnels surface your single biggest drop-off with the revenue blocked by it. It is cookieless and in most cases needs no consent banner; Conclick stores a first-party identifier in localStorage, and whether that requires consent depends on your jurisdiction, so check with your own counsel. Setup takes about two minutes, and there is a 14-day free trial, no card required.

That said: if you are already deep in Google Analytics 4 and your funnel data is clean there, the switching cost is real. GA4 is genuinely more powerful for complex multi-step funnels once you have the data model built. Use the tool that gives you the numbers you need to make decisions, not the one with the most features.

The Discipline That Actually Matters

Fixing funnels is not a project with a finish line. It is a habit. Once a month, look at the funnel, find the biggest leak, and run one focused fix. After three months of this you will have addressed the top three or four leaks in your product. That is usually the difference between a business that limps and one that compounds.

Most founders never do this systematically because the data is scattered across tools, the connection to revenue is indirect, and it is easier to ship features than to stare at a drop-off rate and admit that your current messaging is not working. The discomfort is the work.