One of the contributory causes of landing over-runs is excessive threshold speed. Manufacturers recommend specific additives to the basic Vref. Boeing, for example, recommend adding half the steady headwind component and all of the gust with a total of not more than 20 knots above Vref. Boeing also say that the headwind component additive should be bled off approaching touch-down while maintaining the gust additive. These additives are taken into account in landing distance calculations.
What is often observed are HW additives applied as recommended, but with no real attempt to bleed off the HW additive before arrival over the fence. This may result in a long float. And if the gust additive is maintained, a still longer float or a high touch down speed may occur.
With full gust additive applied right into to the flare, Murphy's Law dictates that the expected lull accompanying the gust will fail to eventuate and you are left with lots of excess speed. When the runway is slippery and a float is allowed to occur to obtain a smooth touch-down, chances are you are risking an over-run. With a gust factor at 90 degrees and the same additive applied, the extra speed is not always dissipated.
While a speed-deficient arrival over the fence is undesirable for several reasons it can be sometimes countered with judicious use of thrust but it takes a deft touch. On the other hand unwanted speed over the fence has been proved to lead to over-runs.
From personal observations on line and in simulators, it is rare to see the half-the-HW component deliberately bled off - mostly people argue it happens naturally at the flare. Maybe so, but not too often. In general, most crews don't worry about it and simply plant the aircraft or go for a smooth landing via the float. Fine if the runway is long and not wet.
I would like to see aircraft manufacturer's have another look at their recommendations on HW and gust additives when conducting manual thrust landings - instead of auto-throttle engaged landings where the Boeing recommendation is to add five knots only to Vref for all landings.
Defining the term "approaching touch-down" where it refers to bleeding off the HW additive, would be helpful. In theory, the free-stream wind gradient starts around 2000 ft and upwards so it is logical to start bleeding off the HW component then -and not leaving it to the flare. Is "approaching touch-down" a specified distance from the runway or a specified height above the aerodrome on final approach?
Accepting some over-runs could be prevented if airspeed control was more precise, manufacturers could consider fine-tuning their advice on the subject of airspeed additives. In short, their strict application may in some cases cause more problems than they are meant to fix.