tl;dr:
An ideal chemical reaction ends with 0% of its inputs and 100% of its desired outputs. If it requires energy input, this occurs immediately. If it releases energy, this occurs on a time scale slow enough that it doesn't blow up, melt down, or otherwise adversely affect the plant, but still rapidly enough to be profitable.
In the long run, the actual amounts at equilibrium will correspond to the energy differences between the inputs and outputs: say we can leave the system alone, and wind up with 10% inputs and 90% outputs; this balances because there are equal numbers of likely transitions forward from the small number of remaining inputs, and unlikely transitions backward from the the large number of current outputs, so even though individual reactions go both ways all the time (slogan: conversion is stochastic), in bulk we don't observe any change. (As we want change, a simple trick is to continually mechanically add inputs and take away outputs, so the system never equilibrates.)
In the short run, how quickly an out-of-equilibrium ratio goes towards equilibrium (the reaction rate) depends upon the energy difference to some intermediate state (the activation energy), which will be more energetic than either input or output. By increasing or decreasing temperature (the going exchange rate between entropy and energy) we can speed up or slow down this rate, because having sufficiently energetic intermediates becomes more or less likely.
Almost all commercially interesting reactions occur in multiple steps. This means that (a) the outputs of one step may be chemically removed as inputs to the next step, (b) in the ideal case we also have 0% of these intermediates but in reality we will always have "work in progress", and (c) the rate limiting step is the one whose activation energy is the greatest; making anything before it faster is pushing on a string, and making anything after it faster is getting blood from a stone.
Finally, note that I said "desired outputs" above. There are often many reactions possible, and again, the ratio between outputs one wants and outputs one doesn't want (eg, in the CS case we call an eigenstate we want "termination" and an eigenstate we don't want "deadlock") depend upon the energy differences between the outputs and between the intermediate states leading to them.
Profitable reactions have multiple reaction steps, but not too many, because each step raises the percentage of undesired outputs (where remaining input is just a degenerate case of undesired output). If we must take 13 95% effective steps, we only have a 50% overall pathway. Conversely, it is often more effective to add two transformation steps, one to put the input into a form that's very easy to convert exactly to something close to what we want, and one to convert that to the desired output, than to attempt to get an ideal pathway in a single step. (the applications here to representations in CS ought to be obvious!)
As far as HN is concerned, the ideal reaction is taking an idea to FU money, so the ChemE concepts above map to finding a pathway that has as its rate limiting step the slowest people who will trade money for what you have to offer.
Did that help stimulate the imagination?