Dark Social: Where B2B Demand Actually Happens Now
By Saroj Jha · July 17, 2026 · 14 min read
Dark social is the buying activity your analytics can't see: links forwarded in Slack DMs, your framework pasted into a private WhatsApp group, a recommendation dropped in a peer community. Your funnel didn't shrink — it went dark.

The invisible journey
9 of 13
buying touchpoints your analytics never sees
Your funnel didn't shrink. It went dark.
Studies place up to 84% of B2B content sharing in these private channels — meaning for every share you can see on public social, several more are happening where no pixel follows.
And it's getting bigger, not smaller. HubSpot's State of Marketing research found B2B buyers now touch about 13 pieces of content before ever talking to sales — up from 8 a few years ago — and roughly 9 of those 13 touchpoints happen in environments marketers can't directly track: AI assistants, private communities, and dark social. Your funnel didn't shrink. It went dark.
How we got here
The term isn't new — Alexis Madrigal coined "dark social" at The Atlantic back in 2012. For a decade it was a footnote in analytics discussions. Three forces turned the footnote into the main event.
First, the private-channel migration: B2B work conversation moved into Slack, Teams, WhatsApp and Discord, and buying conversation moved with it. The question "what tool should we use for X?" is now asked in a 400-person practitioner community before it's ever typed into Google. Second, the privacy era: iOS changes, cookie deprecation, and link-tracking strippers didn't create dark social, but they demolished the tracking that let marketers pretend the visible slice was the whole pie. Third, AI assistants: a growing share of buyer research now happens inside ChatGPT, Gemini, and Copilot conversations — zero-click, zero-referrer, and invisible to your analytics in exactly the way a Slack DM is.
The result, plainly: the majority of your pipeline's formative moments now happen where you cannot watch. The only strategic question is whether you design for that reality or keep optimizing the visible minority.
Why this breaks the playbook you were taught
The last decade's demand machine ran on a simple loop: gate a PDF, count the form fills, call them leads, hand them to sales. That loop assumed the buying journey was visible. In LinkedIn's B2B Marketing Benchmark research, 73% of B2B marketers report declining performance from tactics that worked just a few years ago.
Run the arithmetic on a typical gated asset: strong landing pages convert a small fraction of visitors, so for every hundred readers who wanted your thinking, a handful traded an email for it — and the other ninety-plus either bounced or found an ungated version from a competitor. You suppressed distribution among exactly the people most likely to forward it.
There's a second failure mode: the metrics themselves now lie by omission. A growing share of high-intent buyers never becomes a trackable lead at all — they lurk, they read, they ask their community, and then they arrive as "direct traffic" already 80% decided. Teams then defund the thing creating demand and double down on harvesting it.
Here's the reframe: your content is being distributed for free by your most trusted salespeople — your readers. A VP forwarding your article to three colleagues is doing warmer outreach than any SDR sequence. The job isn't to track it. The job is to be worth forwarding.
The dark-social demand playbook
1. Ungate your best thinking. The working rule: ungate ideas, gate assets. Publish frameworks, teardowns, and points of view openly, because their job is to travel. Gate the things that fairly warrant an email — working tools, templates, calculators. And make your best content copy-pasteable: a framework that survives being pasted into Slack gets pasted into Slack.
2. Make the founder the channel. Forrester projects that by end of 2026, two-thirds of B2B content will come from decentralized, expert-led sources. What works: two to three posts a week from the personal profile, rotating three archetypes — the earned lesson, the contrarian take with receipts, and the useful artifact. Let the company page echo a day later; never lead with it.
3. Show up where the forwarding happens. Niche Slack and Discord communities, sub-1,000-listener podcasts, curated newsletters. These don't scale, which is exactly why they work. Pick two or three communities where your actual buyers live, spend the first month only answering questions with zero links, and let your profile do the selling.
4. Ask the only attribution question that sees in the dark. Add one field everywhere it fits — demo form, signup flow, first sales call: "How did you hear about us?" Use a free-text box, not a dropdown; dropdowns teach people to pick "Google" when the true answer is "someone in a Slack group linked your budget template."
5. Measure in aggregate, not in clicks. Watch the signals dark social does move: branded search volume, direct and unattributed traffic, self-reported attribution mix, inbound quality. For the budget question specifically, aggregate methods like marketing mix modeling sidestep tracking entirely.
The 90-day dark social plan

Weeks 1–2 — instrument the dark. Add the "How did you hear about us?" free-text field to every form and the sales script. Add a Source column to deal tracking. Audit gated content: ungate the thinking, keep the tools gated.
Weeks 3–6 — light the founder channel. Commit to the 2–3×/week cadence with the three post archetypes. Reply to every comment. Repost the best performer to the company page a day later.
Weeks 7–12 — enter the rooms. Join two or three communities where buyers live; give-first for thirty days. Pitch two niche podcasts with a specific take. At day 90, read the evidence: self-reported attribution mix, branded search trend, inbound quality.
Put it to work
The Startup Marketing Plan includes the channel scorecard for two-channels-not-six decisions — and if you want budget math that survives a world without tracking, start with MMM-Lite and its companion piece on incrementality testing.