Funnels that actually converge
Most funnels leak because the steps are wrong, not because the product is. Here is how we define steps that map to real intent instead of lorem-ipsum guesses.
Every team we talk to has a funnel chart, and almost every one of those charts is measuring the wrong thing. The classic mistake is defining steps by page views — visit, pricing, signup — when the user's actual journey is dolor sit amet: nonlinear, repeated, and full of back-buttons. A funnel built on page views converges on confusion.
The fix is to define steps as intent events, not navigation events. "Viewed pricing" is navigation; "compared two plans for more than ten seconds" is intent. In Vertex we call these consectetur events: the small, deliberate actions that only a genuinely interested user performs. When your funnel steps are intent events, the drop-offs stop being noise and start being answers.
A second, quieter failure mode is window size. If your funnel window is a single session, you are silently excluding everyone who thinks overnight — which for a paid product is most people. We default funnel windows to seven days precisely because the tempor incididunt between "heard about it" and "paid for it" is rarely measured in minutes.
Once steps and windows are honest, convergence becomes a design tool. You can ask sharper questions: does the invite step convert better when it appears before or after the first chart renders? Does the checkout step leak more on the plan picker or the card form? Teams at Loremly and Ipsumatic rebuilt their onboarding around exactly these questions and watched step-two completion climb double digits.
The uncomfortable summary: a funnel that never converges is not a broken funnel, it is an honest one. It is telling you the steps you chose do not describe how anyone actually buys. Change the steps, keep the honesty, and the chart will start pointing at work worth doing.