Lay out every stage from first touch to closed revenue with a real conversion rate at each step, then compare those rates against your own historical baseline or comparable businesses at your stage — the stage with the sharpest drop relative to baseline is usually where the leak is, not necessarily the stage with the lowest absolute number. A funnel that converts visitors to leads well but leads to customers poorly points at sales or offer fit; the reverse points at targeting or top-of-funnel quality.
Segment by source before concluding anything — a leak that looks funnel-wide is often really one bad channel dragging the blended average down.