A vendor added one survey question. 31% of 'direct' leads were actually dark social.

A vendor added one survey question. 31% of 'direct' leads were actually dark social.

The question: how much pipeline does dark social (sharing via DMs, Slack, email — invisible to analytics) actually hide?

The case: a cybersecurity vendor appended a single 'How did you first hear about us?' field to its demo form and reconciled answers against 'attribution' (the analytics model assigning credit to channels) for 6 months. Sample: 412 self-reported responses.

Three findings:
— 31% of leads that analytics labeled 'direct/unknown' named a specific LinkedIn post, podcast, or peer DM — channels with zero tracked clicks.
— LinkedIn was named 2.7x more often than its click-attribution credited it.
— Leads from named dark-social sources closed at a higher rate (38% vs 22%) than tracked-paid leads.

Limitations: self-reported memory is noisy and recency-biased; people overweight what they saw last. Single vendor.

What it means for B2B: your dashboard systematically under-credits the channels that actually persuade, because persuasion happens in private.

Bottom line: one form field recovered a third of 'unknown' demand — cheap instrumentation, large blind spot.