Attribution
Dark social
Also called unshareable share.
Dark social is traffic from private shares, such as messaging apps and email, whose referrer is stripped, so it often lands as direct. The sharing happened, but your tool cannot see where it came from.
How it is measured
You cannot measure it directly. You infer it from long, clean URLs that show up as direct entries, since people rarely type those. Landing page patterns and the share of direct on deep pages show the scale.
A way to confirm is to add share links with a tag on your own pages and compare tagged share traffic with the direct landings on those articles.
Worked example
A travel blog's article 'Three days in Lisbon' draws 1,800 direct-labeled sessions in a week. Nobody types that long address. The share button data shows only 190 clicks.
The owner adds a copy-link button that appends a share tag, and next week 520 of those direct sessions arrive with it. The remaining 1,280 are still dark, but the scale of private sharing is now visible.
How it differs
Dark social is private sharing with hidden origin. Direct traffic is the bucket where it hides. Direct also contains bookmarks and typed visits, so not all direct is dark social.
Common errors
Calling all direct traffic dark social. Ignoring email apps that strip referrers. Not tagging your own share buttons. Assuming social teams own it. Counting it as zero because you cannot see it.
In practice
Look at your top direct landing pages. If they are deep article pages, assume private sharing and put tagged share links on them. Do not credit the home page with those visits.