Did some quick maths with some simplistic assumptions about accels and decels and that "rule of ten" is pretty good. It actually looks like it UNDERESTIMATES the runway exit speed for small values - it looks about right for 5-6 knots, for reasonable V1 values. For one or two knots, the multiple is more like 20, though. And for ten or more knots delay the factor drops to 7 or lower.
If I use the 40/70 break points for hazards, then any delay of more than, say, 4 knots is hazardous, and anything over 9 or 10 knots is catastrophic. And, as OORW astutely notes, that's assuming a relatively "friendly" overrun area...
Assuming a (lowish) takeoff T/W of, say, 0.25'g', that's an accel of about 5knots/sec. A light weight could see that nearly double. So those speed increases equate to about a second of delay.