What Is a Conversion Funnel?

A conversion funnel reduces a multi-step process to a set of numbers: out of a hundred page visits, how many reach the next step, and exactly where most people drop off. It's a quantitative snapshot of the process, not a description of what the user is thinking at any given moment.

The most common example is the e-commerce funnel: product page visit, add to cart, checkout started, purchase completed. Each step is built on a measurable event, and the traffic at the first stage becomes the base against which the percentage reaching the last stage gets calculated.

It shouldn't be confused with the customer journey. A funnel measures how many users move through a specific, measurable process with predefined steps. The customer journey describes a person's full path across multiple channels, including moments that don't always reduce to an analytics event, like reading a review on another site before coming back to buy.

The three models coexist without replacing each other: the AIDA model explains why someone moves forward, the customer journey documents where that person goes, and the funnel counts how many make it to the end.

The technical foundation of a funnel is web analytics with event tracking: each stage corresponds to a logged action, such as a page view, a click on "add to cart," or a confirmed purchase. Tools like Google Analytics let you define up to ten steps in a single funnel, compare user segments against each other, and set a maximum time window between one step and the next.

Drop-off is calculated per stage, not just for the funnel as a whole: if a thousand people view a product page and seven hundred add something to the cart, drop-off at that step is thirty percent. That figure, repeated stage by stage, points to where most people are actually lost, something a single overall conversion rate doesn't reveal on its own.

Google Analytics also distinguishes between open and closed funnels: a closed funnel only counts users who enter at the first stage, while an open one also counts those who land directly on a middle stage, for example someone arriving straight at the product page from an ad. Picking the wrong setting distorts the real drop-off rate at each step, so it's worth checking that configuration before drawing conclusions from the numbers.

Google Ads applies similar logic in its campaign reports, with funnels running from ad impression to final conversion, useful for comparing performance across different paid traffic sources.

A funnel points to the stage with the biggest drop-off, the real bottleneck in the process, and keeps optimization effort from spreading evenly across steps that already work fine. If the biggest loss happens between cart and checkout, adjusting the call to action there or simplifying the form usually pays off more than tweaking the product page.

It also helps decide where to invest at the top of the process: a well-placed lead magnet can grow the volume entering the funnel, but that only makes sense once the following stages are already converting normally.

Even so, a funnel remains a simplified model, and it's worth treating it that way. It assumes a linear path, step one, step two, step three, when real behavior rarely follows that order: some users jump straight into checkout from an ad, others visit the product page on several different days before deciding. It's the same limitation carried by the AIDA model: useful as a reference framework, incomplete as an accurate picture of actual behavior.

• Define each stage around a specific, measurable event, not an intent that's hard to track.
• Compare drop-off stage by stage, not just the overall conversion rate from start to finish.
• Check the funnel separately by traffic segment, since drop-off on paid campaigns can differ sharply from organic search.
• Prioritize optimization at the stage with the biggest loss before touching the rest of the process.
• Set a reasonable time window between stages, so someone who buys again days later doesn't get counted as a drop-off.
• Cross-reference the funnel with qualitative data, like session recordings, to understand why users are lost at a specific stage.

• Treating the funnel as if it describes all user behavior, instead of one specific, bounded process.
• Defining stages too broadly, mixing different actions together and hiding where drop-off actually happens.
• Ignoring users who enter the funnel at a middle stage instead of always at the start.
• Comparing the conversion rate of two funnels without accounting for the fact that they measure different processes.
• Optimizing the stage with the smallest drop-off because it's the easiest to improve, instead of the one losing the most volume.