Start with what is between the card and the computer. A microSD in an SD adapter has two sets of contacts and a plastic switch that lies; try it without the adapter in a reader with a microSD slot. A full-size card in a laptop's built-in slot may be meeting a slot that does not do UHS-II, or does not do SDXC at all on an older machine; try a current USB reader. A Mac that refuses to mount a card may be refusing exFAT on an old system or ext4 from a games console; look in Disk Utility, where the device appears even when the volume does not. A card the camera reads and the computer does not is a reader or a file-system problem, not a card problem, and the camera's USB connection will often copy the files out itself. One good reader on one good machine is enough. Twenty readers is not more information; it is twenty more chances for a weak card to lose a block.
Then look at what the computer does see. On Windows, Disk Management or Device Manager will show one of three things. Nothing at all means the card drew no current and answered nothing: a dead controller, a short, or a card so cracked its contacts are gone. A device with no media, or 0 bytes, means the controller is alive and answered the host but cannot find its own translation table in the flash, which is the same fault in a less final form. A device with a capacity, and a volume Windows cannot read, is the format message, and a different page. On a Mac, Disk Utility shows the same three states, with a device greyed out where Windows would say no media.
When it is the card, the data has not gone anywhere. The flash chips hold what they held; it is the controller, the small processor that answered the computer's questions and translated addresses into flash locations, that has stopped. On a CompactFlash card and an older or larger SD card, the flash is in separate chips on a small board, and they are desoldered and read on a programmer. On a microSD, and on most modern SD cards and small USB sticks, the controller and the flash are one sealed block, and the flash is reached through the test pads the factory used, under the coating on the back: pads identified against a pinout database or by probing, a needle adapter fitted, the raw flash read out. Either way the dump is scrambled, has error-correction data interleaved, and is laid out in the controller's own page and block order, and the bench's software undoes each of those to rebuild the card as the computer would have seen it. How a memory card is recovered, from the reader to the monolith goes through it step by step.
Two things are worth knowing before you send it. A card that was not seen because its contacts were dirty or corroded, after a swim or a washing machine, is cleaned and read, and is the lower band. And a card that has been through many readers and phones in the hope that one would see it may have been powered and failed thirty times, each one a chance for the controller to write a bad block to its own table; the card that was tried once and sent is in better shape than the card that was tried everywhere.