Colour spaces explained for people who sell things
Colour management is the part of image handling most sellers skip, and mostly they get away with it. Where they do not, the symptom is expensive: a return, or a review saying the item is not the colour shown.
The underlying idea takes about five minutes and is worth having.
The basic problem
A pixel is three numbers. Nothing about them says which colour they mean.
RGB(200, 30, 40) is a red, but which red depends on what the numbers are measured against. Say "200 out of 255 towards the reddest red" and the obvious question is: the reddest red of what?
A colour space answers that. It defines exactly which physical colours count as fully red, green and blue, plus a white point and a curve describing how the stored numbers relate to actual light.
The four you will meet
sRGB. The default for the web, defined in 1996 around the capabilities of monitors of the time. The smallest gamut of the four. Every marketplace serves sRGB, and every device displays it correctly. Its limitation is real: it cannot represent the most saturated reds, greens and cyans that a modern screen can show.
Adobe RGB. Wider, particularly in greens and cyans. Designed with print in mind, since CMYK printing can reach greens sRGB cannot.
Display P3. Wider than sRGB in reds and greens, the same in blue. What Apple screens use, and what iPhones capture by default. This is the one most likely to be sitting in your files without your knowing.
ProPhoto RGB. Enormous, so wide it includes colours the human eye cannot see. Used as an editing space where you want headroom, never as a delivery format.
What goes wrong
The failure has one shape: numbers from one space interpreted as another.
P3 numbers read as sRGB come out less saturated than intended, because P3's 255,0,0 is a more saturated red than sRGB's, and reading it as sRGB just gives you sRGB's red. Everything shifts inward. Saturated colours suffer most; neutrals are unaffected.
That is the diagnostic. If your greys and whites are right and your reds are dull, it is a profile problem, not exposure.
The reverse happens too. sRGB numbers read as Adobe RGB come out over-saturated, colours look punchy and wrong, and skin tones in lifestyle shots go orange.
Why it matters commercially
For most product categories this is an aesthetic issue. For a few it is a returns issue.
Apparel and textiles. Colour is a purchase criterion. A navy that photographs as royal blue produces returns.
Paint, dye, thread, yarn, cosmetics. The colour is the product.
Anything sold as a matched set. Two items photographed on different days under different white balance look mismatched in the listing even when they match in reality.
For a phone case in three colours, nobody returns anything over a slight shift. For a foundation shade, they do.
Five things worth doing
Convert to sRGB on export, deliberately. Not assign. Convert. Assign relabels the numbers and produces exactly the problem above.
Keep the profile embedded until you are certain the file is sRGB. An untagged wide-gamut file is guaranteed to be misread by something.
Set white balance manually and keep it fixed across a product range. Auto white balance drifting between shots is a bigger practical source of colour inconsistency than gamut ever is.
Check on a second device, ideally one that is not Apple. A P3 screen renders an untagged P3 file roughly correctly by accident, which is why the problem so often surfaces only when a customer complains.
Photograph a colour reference in one frame per session if colour accuracy really matters. Even a sheet of known white paper in the corner of one shot gives you something to correct against later.
What you cannot fix
Screens differ. A shopper on an uncalibrated laptop at 60% brightness with a blue-shifted panel sees something different from you, and no amount of colour management on your side changes that.
The goal is not perfect reproduction on every device, which is unattainable. It is to remove the errors you control, so that what remains is the shopper's screen rather than your export settings.
Where colour genuinely decides the purchase, the honest addition is a description: "deep navy, closer to the second photograph on most screens". A sentence costs nothing and prevents returns that a colour profile cannot.