Paper is the biggest single cause
Ink is transparent. What you see is ink over paper, so a change in the paper's shade, coating or absorbency changes the colour even when the ink is identical.
A stock substitution - often made in good faith when the specified paper is unavailable - is the most common reason a repeat run looks different.
The proof was on a screen
Screens are backlit and vary by device. Approving colour on one and then judging the print against a different one is comparing two things that were never the same.
For anything that has to match, approve on a physical proof and keep it.
The job moved between suppliers
Two presses, two ink sets, two operators, two houses' idea of what "the same" means. Nothing was done wrong; the result is still different.
This is the strongest practical argument for keeping a repeating job with one press.
The build was RGB, not CMYK
Some colours that exist on screen simply cannot be made with process inks. If artwork is built in RGB and converted at the last moment, the conversion decides the colour, and different conversions decide differently.
Build in the colour space it will print in, and specify critical brand colours rather than leaving them to be interpreted.
And one that genuinely is variation
Within a single run, some drift is physical - ink density, temperature, humidity. A controlled press keeps it inside a tolerance the eye does not notice. It never gets it to zero, and anybody promising zero is describing something other than printing.
How to keep a repeat matching
Specify the paper by name and refuse casual substitutions. Approve on a physical proof and retain it. Keep the job with one press. Build in CMYK and specify brand colours. Then a reprint matches because it was set up to, not because it got lucky.
What a tolerance actually means
No press prints identically twice. What a controlled press does is keep variation inside a tolerance small enough that the eye does not register it in normal viewing.
That is a real, achievable standard. 'Exactly the same' is not, and a supplier promising it is either misunderstanding the question or hoping you will not notice.
Brand colours deserve to be specified, not described
'Our red' is not a specification. A named colour reference, or a retained physical sample, is. Without one, every reprint is an act of interpretation by whoever happens to be on the press that day.
For a brand that prints regularly, agreeing that reference once is one of the highest-return hours anyone will spend.
Coated and uncoated will never match, and that is physics
The same ink on coated and uncoated stock produces visibly different colour, because uncoated paper absorbs ink and coated paper holds it on the surface.
If a brand appears on both, the honest answer is to approve a separate target for each rather than to chase an impossible match.
Keep a retained sample and date it
One physical piece from every approved run, dated and stored somewhere that is not a drawer full of everything else. It costs nothing and it is the reference every future run is measured against.
Most colour disputes are not really about colour. They are about the absence of an agreed reference, and this is the reference.
