The funnel is a reporting convenience, not a description
Awareness, consideration, decision. It is a useful way to organise a budget and a poor description of what any individual buyer does. Real paths loop, stall, restart months later and frequently skip stages entirely.
That matters commercially because budget gets allocated by stage, and stages that are artefacts of the diagram attract spend they do not deserve.
What the data usually shows
- Most sessions that end in a sale are not the first session. The visitor came back, often several times, and analytics attributed the sale to whichever touch happened to be last.
- The pages that appear in conversion paths are frequently unglamorous — pricing, about, a specific service page — not the campaign landing page the budget was built around.
- A meaningful share of buyers arrive already decided, having done their research somewhere you cannot see. Generative engines have made this larger, not smaller.
- Re-entry points matter more than entry points. Someone returning to compare is closer to buying than someone arriving cold.
What to do instead of mapping a funnel
Find the pages that appear in paths that converted, and the pages where paths ended. Not the pages with the most traffic — the pages that appear disproportionately in each set. That is a different list, and it is usually surprising.
Then ask what a visitor at that page was trying to establish, and whether the page establishes it. Most conversion problems are questions the page did not answer, not persuasion the page failed to apply.
The measurement caveat
Last-click attribution will tell you the final touch mattered most. It always does, by construction. Before restructuring a budget around it, check what the same data says under first-click and linear models — if the answer changes materially, you have learned that the model is doing the deciding, not the data.