On Friday FDA codified a new device lane — the diabetes digital behavioral therapeutic, 21 CFR 880.5735. A lane's value is the traffic it carries. Here are all 846 FDA product-code lanes with meaningful 510(k) traffic since 2016: how busy each one is, and how fast FDA moves the cars through.
Each dot is one FDA product code. Right = more 510(k)s ride the lane. Up = FDA's median review takes longer. Dot size tracks traffic. The red dots are the 13 digital-behavioral-therapeutic lanes — the family the new diabetes classification just joined. Hover anything; filter by specialty; then drag the sample-size floor and watch the "fastest lane" headline change.
The empty 78% is invisible here. A log axis that starts at 3 hides 5,476 lanes that carried nothing for a decade. The modal FDA lane is not LLZ — it's silence. A new classification like QXC starts there, and most never leave.
Small-n medians lie. At the floor of 3, the "fastest lane in FDA" is a 3-clearance fluke (a lane's median can be set by a single resubmission). Drag the floor to 20+ and the extremes converge toward the boring middle (~90–160 days). Any vendor deck quoting a lane's "typical clearance time" off a handful of 510(k)s is quoting noise.
What's not counted: De Novo grants (the lane-creating pioneer devices), PMAs, and non-SE decisions. Review days measure FDA's clock, not your build — a 517-day outlier usually means additional-information cycles, not a slow reviewer.