This guide gives a clear three-way distinction, and a short example of how their relative weight shifts with batch size.
Setup time
Setup time is the time needed to get the machine, workholding, tools and process physically ready to run a given operation: mounting and aligning the workholding, loading and verifying tools, and any other preparation that happens before the machine starts cutting. Setup time is typically the responsibility of a machine operator or the person doing work preparation, and it is incurred at the start of an operation (and often again each time the workholding setup changes).
Programming time
Programming time is the time needed to create the CNC or CAM program, and any related process preparation, for a given part or operation. This is usually a distinct role, a CNC programmer, separate from the person doing work preparation, and separate again from the machine operator. Programming time is incurred once per program (though a program may need rework if the part or process changes), not once per part produced.
Cycle time
Cycle time is the time the repeating machining cycle itself takes to run, once setup and programming are already done. This is the time that repeats for every part in a batch. Cycle time is not predicted the same way in every manufacturing context: what dominates it depends on the process, the batch size, the machine, the number of workholding setups, and the part's own requirements. A milling operation, a turning operation and an EDM operation, for instance, have fundamentally different variables driving their cycle time, because material is removed in fundamentally different ways.
Why these three should never be one number
Setup time, programming time and cycle time behave differently as a function of batch size, have different owners, and are affected by different variables. A company that tracks them as a single undifferentiated "production time" loses the ability to answer basic, commercially important questions: is this job expensive because the part takes a long time to cut, or because it needs an unusually complex setup? Would quoting a slightly larger batch meaningfully reduce the price per part, or not? Answering those questions requires the three components to stay separate, from estimating through to the actual times recorded during and after production.
A worked example: one-off versus large series
Consider a part that needs, roughly, 90 minutes of setup time, 60 minutes of programming time, and 10 minutes of cycle time per part.
For a single part (a batch of one), the total time is 90 + 60 + 10 = 160 minutes. Setup and programming together make up the large majority of that total (150 of 160 minutes, roughly 94 percent). The cycle time itself is almost incidental to the total cost of the job.
For a batch of 100 identical parts, the setup and programming time is still incurred once (150 minutes total), but the cycle time is incurred 100 times (1,000 minutes total). The total time is 1,150 minutes, and cycle time now makes up the large majority of that total (roughly 87 percent). The same part, with the same setup and the same program, has a completely different cost structure once the batch size changes.
This is why a repeat order is not automatically priced the same way as a one-off, and why a large series can often be quoted more competitively per part than a small one, not because the part becomes easier to make, but because the fixed setup and programming cost is spread over more parts. It is also part of why a genuine repeat order (the same part, the same revision, the same route) can sometimes reuse an existing program and setup knowledge, while a revised part may need some or all of that preparation redone, even if the geometry looks similar.
A note on measurement and learning
The setup time, programming time and cycle time recorded during actual production are valuable, but only once it is clear the recorded values are representative of normal conditions. A setup that took unusually long because of a tooling problem, or a cycle time affected by an unplanned stoppage, is a legitimate production outcome, but not necessarily a good input for the next estimate on a similar part. Separating these three time components clearly is also what makes it possible to tell, later, exactly which part of an estimate was off and why, rather than being left with a single number that was simply wrong.