Real sequential funnels — each stage is the same people narrowed down, not four unrelated metrics on one screen. Where a channel isn’t measurable yet, LeadForge says “not tracked yet” instead of inventing a number.
Reached, engaged, captured, qualified — the same cohort followed through, so a drop between two stages is a real leak you can go and fix. No blended averages, no metrics from different populations stacked next to each other.
The Agent has real access to your funnel performance, A/B results and attribution — for answering questions, not taking unsupervised action. It analyses; you still decide what to do about it.
Variants are compared on leads produced, and the winner becomes the default for the next draft.
Split the same funnel by audience segment to see which persona is worth writing more for.
Export the funnel and attribution view when someone asks what the content is doing.