When I was in college, we tried to calculate drag coefficients for a car using the coast-down method. The main problem is that the measurements aren't very accurate. For example, on my speedometer (a cheapy), displayed speed lags a second or so behind (it allows several wheel rotations before recognizing a change) and then also, speed may change in fairly large increments- so it may show 20.1 and 20.4 without showing 20.2 and 20.3 in between. The result is your measured deceleration is not very accurate.

With the car test, this was done in west Texas, which is noted for wind, and we did the test in both directions in hopes of averaging out the wind factor, but that throws another big unknown into it. Also, you don't normally have a direct headwind or tailwind, and the effects of a quartering wind are not normally figured in drag calculations, so it just makes the whole approach that much more approximate.

The method also assumes you know what you and your bike weigh.

It would likely be about as useful to use one of the online calculators.