It was discovered very early in the history of chemistry that the speed at which a chemical reaction happens and the amount of reactants and products left once a reaction reaches a steady state are two very different things. The first refers to the kinetics of the reaction, and the second to the reaction's equilibrium. Although there are empirical principles that suggest the two are related, such as the Hammond Postulate, there is no necessary relation between them.
One of the key differences between kinetics and equilibrium is that kinetics depends on the concentration of reactants, but equilibrium (more specifically, equilibrium ratio) does not. The fact that the speed of a reaction depends on reagent concentrations often causes confusion with regard to the units of the rate constant, an arbitrary constant that relates the concentrations of species in a reaction to the observed rate of the process. This confusion has puzzled many a chemistry student over the years.
The rate of a reaction is always expressed in units of concentration per unit time, often molarity per second or M/s. The dependence of rate on concentration varies depending on the species involved in the rate-determining step of the reaction and the steps leading up to these species. The dependence may be quadratic, cubic, or of even higher order if multiple molecules are involved in the rate-determining step. But in order for both sides of the rate law to give M/s, any concentration terms beyond the first must be canceled out by a concentration term in the denominator of the rate constant! Consider the one-step SN2 reaction of methyl tosylate with methoxide:
CH3OTs + -OCH3 --> CH3OCH3 + -OTs
This is a one-step process, so we know that the step shown is the rate-determining step. One molecule each of the reactants enters into the transition state, so the rate law is:
Rate = k(CH3OTs)(-OCH3)
Given that the rate must be expressed in M/s, we see that the rate constant must take on the bizarre units of M-1s-1. Does such a rate constant have any physical meaning? For unimolecular processes, one can consider the rate constant (with units of 1/s or, for the physicists in the audience, Hertz) as the frequency at which the (single) compound experiences the process. For bimolecular reactions, such as the SN2 process outlined above, the rate constant still represents the frequency of the process, but must also take into account the tendency of the molecules to "bump into" each other. A large rate constant for a bimolecular process represents both the tendency of the molecules to collide and the tendency of the molecules to react when they do collide. In a rudimentary way, the "per molar" term of the rate constant can be thought of as measuring this collisional tendency: how often does a molecule of species A bump into a molecule of species B per molar of species A, at constant concentration of B? For a termolecular step, whose rate constant is expressed in M-2s-1: how often do molecules of species A and B collide simultaneously with species C, at constant concentration of B and C?
The collisional properties of molecules of comparable molecular weight are usually pretty similar, so comparing two rate constants usually gives us information about how often molecules react when they do collide--dividing one rate constant by another often "cancels" collisional terms. This is why we can usually talk about reactions in "molecule-on-molecule" terms: we don't have to worry about how fast the molecules are colliding with each other, just the molecular properties that determine if they should react when they do collide. The fact'll make you appreciate how complicated chemistry could be.