The Attention Landscape · established-phenomenon-flagged-statistic evidence
Dark Social at 13: What We Actually Know (and Don't) About Untrackable Sharing in 2026
"Dark social" is the term, coined by The Atlantic's Alexis Madrigal in 2012, for sharing that happens through private, untrackable channels, email, text, and direct message, rather than the public shares an analytics tool can attribute to a named platform. That same year, ad-tech firm RadiumOne's "The Light and Dark of Social Sharing" study found roughly 69 percent of social sharing happened through these dark channels, about three times the volume attributed to Facebook and all public channels combined. That 69 percent figure is now fourteen years old, drawn from a single vendor study, and should not be repeated as a current statistic; most present-day claims that "dark social is still around 70 percent" simply re-cite the same 2012 number without new measurement. What has not gone stale is the underlying phenomenon, and the case that it matters more today, not less: messaging apps, the primary dark-social channel, now carry more monthly reach than any public social feed, with WhatsApp alone at 3 billion-plus monthly users. This piece separates what is genuinely established from what is an outdated statistic still being passed off as current.
What "dark social" actually means
The term describes a specific, structural blind spot in web analytics, not a conspiracy or a technical flaw. When someone shares a link by copying and pasting it into a text message, an email, or a private chat, the person who clicks that link typically arrives at the destination site with no referrer information attached. Standard analytics tools, unable to see where the click came from, classify that visit as "direct traffic," the same bucket used for someone typing a URL straight into their browser.
That misclassification is the entire mechanism. A visit that actually originated from a friend's private recommendation gets counted identically to a visit from someone who already knew the business's exact web address. The two sit at opposite ends of what they tell a business about a visitor, one an earned, trusted referral and the other a pre-existing brand relationship, and conventional analytics cannot tell them apart.
Where the term and the number both came from
Alexis Madrigal, then an editor at The Atlantic, coined "dark social" in a 2012 piece titled "Dark Social: We Have the Whole History of the Web Wrong," describing exactly the referrer-blindness mechanism above. In the same year, the ad-tech firm RadiumOne published "The Light and Dark of Social Sharing," a study that attempted to quantify the phenomenon and found that approximately 69 percent of all social sharing occurred through these private, untrackable channels, roughly three times the sharing volume attributed to Facebook and all public social channels combined.
Both the term and the number, in other words, originate from the same year, 2012, and the number comes from a single vendor-sponsored study. That is not a reason to dismiss it as false; it is a reason to be precise about what it establishes. It established, credibly, that dark social was already the majority channel for sharing fourteen years ago. It did not establish, and cannot establish on its own, what the figure is today.
Why the 69 percent figure should not be quoted as current
The 69 percent figure does not hold up as a current statistic. A search for "dark social statistics" turns up article after article citing "around 70 percent" as though it were a fresh, ongoing measurement. Nearly every one of those citations traces back, directly or through several layers of secondary re-citation, to the same 2012 RadiumOne study. No comparably rigorous, recent measurement, from 2024, 2025, or 2026, was located for this piece.
That is a genuine, structural gap in the public data. Digital media consumption has changed enormously since 2012: messaging-app usage has multiplied, short-form video did not yet exist at scale, and an entirely new AI-answer layer has emerged. A statistic measured before any of that happened cannot honestly describe the present composition of dark social, even if the underlying phenomenon it once measured is, if anything, larger today. The correct statement is: dark social was measured at roughly 69 percent of sharing in 2012, and no rigorous recent figure exists to update that number. Any article stating "69 percent" or "70 percent" as a 2026 fact is repeating a fourteen-year-old number without saying so.
The case that the phenomenon is larger today, even without a fresh number
While the precise percentage cannot be honestly updated, the scale of the channels that carry dark social sharing can be, and that scale has grown substantially. Messaging apps, the dominant dark-social vehicle, now carry more monthly reach than any single public social feed. WhatsApp reports between 3 billion and 3.3 billion-plus monthly active users, depending on the tracking window. WeChat, dominant in China, reports approximately 1.4 billion. Meta's Messenger crossed back over 1 billion monthly actives in 2025. Telegram's founder disclosed roughly 1 billion monthly actives in March 2025. Snapchat reported roughly 932 million monthly actives in Q2 2025.
None of these figures directly measures the share of sharing that is dark, and citing them as if they proved a specific percentage would repeat the same error this piece is correcting in the RadiumOne figure. What they do establish, credibly and independently, is that the infrastructure through which dark social sharing happens has scaled dramatically since 2012, when Facebook itself was the dominant comparison point. A structurally larger set of private channels carrying more total monthly reach than any point in 2012 is reasonable grounds to argue the underlying phenomenon has grown, even without a matching updated percentage.
What this means for a business reading its own analytics
The practical consequence for any business watching a "direct traffic" line in its analytics dashboard is that a meaningful, likely majority, share of that line is not people who already knew the exact URL. It is dark social: word of mouth, private recommendations, group-chat mentions, and text-message shares that a standard analytics tool cannot distinguish from a cold, unattributed visit. That misclassification quietly erases the evidence of a business's strongest and cheapest acquisition channel, genuine referral, from its own reporting.
This is precisely the kind of gap a demand-side attention measurement, one that samples what buyers are actually seeing and sharing rather than only what a supply-side traffic log can attribute, is built to address. A traffic log can only ever report where a visit says it came from. It cannot report the private conversation that actually prompted the visit.
What would need to exist to close this gap
This topic requires naming the actual research gap plainly: no recent, rigorous, methodologically disclosed dark-social measurement study, comparable in quality to RadiumOne's 2012 work, currently exists in the public record. That is a genuine opening for primary research that measures the phenomenon as it actually behaves today, across a set of channels RadiumOne's researchers could not have anticipated.
Until such a study exists, treat dark social as a well-evidenced, real, and probably growing phenomenon, and treat any specific current percentage attached to it as an unverified, outdated claim carried forward from 2012, regardless of how many places repeat it.
The evidence
Key findings, with their sources
-
RadiumOne's 2012 study found approximately 69% of social sharing occurred through private, untrackable channels, roughly three times the volume attributed to Facebook and all public channels combined.
established RadiumOne, "The Light and Dark of Social Sharing," 2012.
-
The 69% figure is a single 2012 vendor study and should be treated as a historical baseline, not a current statistic; no comparably rigorous 2024-2026 measurement was located.
contested Cross-referenced against current secondary sources, all tracing back to the 2012 RadiumOne study.
-
WhatsApp reports 3 to 3.3 billion-plus monthly active users; WeChat approximately 1.4 billion; Messenger back over 1 billion in 2025; Telegram approximately 1 billion (March 2025); Snapchat approximately 932 million (Q2 2025).
established Statista, "Most popular messaging apps 2025"; Business of Apps, "Messaging App Revenue and Usage Statistics 2026"; company disclosures.
Reference
Glossary
- Sharing that happens through private, untrackable channels such as email, text messages, and direct messages, which conventional web analytics cannot attribute to a specific source and instead classifies as direct traffic.
- Direct traffic
- The analytics category for visits with no identifiable referrer, used both for genuinely direct visits (typing a URL) and for dark-social visits, which analytics tools cannot tell apart.
- Referrer
- The data a browser sends to a destination site indicating which page a visitor came from. Private-channel shares typically strip this data, which is the technical root cause of the dark-social measurement gap.
Straight answers
Frequently asked questions
What is dark social?
Dark social is sharing that happens through private, untrackable channels, email, text messages, and direct messages, that conventional web analytics cannot attribute to a source. It gets misclassified as "direct traffic," the same bucket used for someone typing a URL straight into their browser.
Is the 69% dark social statistic still accurate in 2026?
No, or at least it cannot be confirmed as accurate. That figure comes from a single 2012 study by RadiumOne. No comparably rigorous, recent measurement exists in the public record, so any source citing "70% of sharing is dark social" today is repeating a fourteen-year-old number, usually without saying so.
If the statistic is outdated, does that mean dark social matters less now?
The opposite case is better supported. Messaging apps, the main dark-social channel, have grown enormously since 2012; WhatsApp alone now reports 3 billion-plus monthly users. The infrastructure carrying dark social sharing is larger today than at any point when the original 69% figure was measured.
How would a business know how much of its own traffic is dark social?
Standard analytics cannot separate genuine dark-social visits from truly direct ones, since both lack a referrer. Estimating the split requires deliberate methods such as UTM-tagged private shares, post-purchase attribution surveys, or per-client referrer-pattern analysis, none of which a default analytics setup performs automatically.
Why does dark social matter for AI-search visibility specifically?
Both problems share the same root cause: a large and possibly growing share of how buyers actually discover a business happens outside the channels conventional supply-side analytics were built to measure, whether that is a private message or a synthesized AI answer with no click attached.
Provenance
Sources
- Alexis C. Madrigal, "Dark Social: We Have the Whole History of the Web Wrong," The Atlantic, 2012 (established, origin of the term)theatlantic.com
- RadiumOne, "The Light and Dark of Social Sharing," 2012 (established as a dated baseline; the specific 69% figure flagged contested for present-day use)
- Statista, "Most popular messaging apps 2025" (established)statista.com
- Business of Apps, "Messaging App Revenue and Usage Statistics 2026" (established, vendor-aggregated)businessofapps.com
Every figure above is attributed to a real, dated source and tagged with its evidence tier. Where a claim could not be verified to a primary source, it is not stated as fact.