Privacy
Fingerprinting
Also called device fingerprint.
Fingerprinting is building a likely-unique signature from device and browser traits instead of storing an ID. Fonts, screen size, graphics output, and language settings are typical inputs.
How it is measured
A script reads dozens of attributes and combines them into a hash. If the combination is rare, it works like an ID. Test by running the same script on two browsers and comparing outputs.
The strength depends on entropy. A fingerprint that matches 1 in 2,000 browsers is weak. One that matches 1 in 200,000 follows people reliably.
Worked example
A researcher tests a fingerprint script on 3,000 volunteers. 2,100 get a unique value. After a browser update, 400 of them change value, which shows a fingerprint is less stable than it first looks.
The ad vendor using it must now re-match those people by other signals.
How it differs
Fingerprinting derives an ID from traits. Cookieless tracking avoids any lasting ID. Fingerprinting excludes the stored key but still tracks; cookieless excludes both the stored key and the signature.
Common errors
Calling it privacy-friendly. Believing no consent is needed because nothing is stored. Using it silently. Ignoring that browsers now resist it. Treating a fingerprint as stable.
In practice
Check your scripts and vendors for fingerprinting. Ask them directly. Remove any that cannot say no.