The short version
Dark social is sharing that happens in private channels analytics can't see: Slack and WhatsApp messages, email forwards, DMs, podcasts, communities and, increasingly, AI assistant answers.
In B2B it's often where buying decisions start. When those buyers later visit your site, the visit shows up as direct traffic or branded search, and the real source disappears.
You can't track dark social directly, but you can measure it with three methods together: a self-reported source question, tagged links for every referral path you control and trend signals in direct and branded search.
The goal isn't perfect attribution. It's knowing enough to stop cutting the channels that create demand privately.
A founder looks at last quarter's closed deals. Six of the ten show "direct" as the source. On the first sales call, though, those buyers said things like "a friend in our Slack group mentioned you", "I heard your CEO on a podcast" and "ChatGPT listed you when I asked about agencies". None of that is in the CRM.
That gap is dark social. This guide explains what dark social is, where it hides in your data, why it matters more in B2B, the three methods that measure it, how to set up a self-reported survey that works, how to read the signals each month and the mistakes to avoid.
What dark social is
Dark social is any sharing of your content or name through channels that don't pass referral information to your analytics. The term was coined by Alexis Madrigal in The Atlantic in 2012, when he noticed that a large share of traffic arriving as "direct" was really people sharing links privately.
Back then it mostly meant email and instant messaging. For a B2B firm today, it covers:
Private messaging: Slack and Teams channels, WhatsApp groups, LinkedIn DMs and forwarded emails
Communities: founder groups, industry Slack communities, private forums and peer networks
Audio and video: podcast mentions, webinars and conference talks
Social posts read without clicking: a LinkedIn post that makes someone search your name a week later
AI assistants: answers from ChatGPT, Perplexity, Gemini or Claude that name you, which buyers act on later
Offline word of mouth: a recommendation over coffee or at an event
Where dark social hides in your data
Dark social doesn't disappear from your analytics. It turns up under the wrong label. Google Analytics 4 defines direct traffic as visits where users "arrive at your site/app via a saved link or by entering your URL" (Google Analytics Help). In practice, untagged links opened from apps, documents and many private channels often land there too.
What actually happened | What your reports usually show |
|---|---|
A colleague shares your guide in Slack and the buyer clicks | Direct traffic |
A buyer hears you on a podcast and types your name into Google | Branded organic search |
A buyer reads a LinkedIn post, then visits your site days later | Direct or branded search |
A peer recommends you on a call and forwards your link by email | Direct, or "email" if your own emails aren't tagged |
An AI assistant recommends you and the buyer searches your name | Branded search, unless they click a cited link |
The result is that direct and branded search get credit for demand they didn't create, while the channels that did create it look weak. Our guide to last-touch attribution explains why that misleads budget decisions.
Why dark social matters more in B2B
B2B buyers lean on peers because the stakes are high and the options are hard to compare. Much of that research happens before they ever contact a vendor. 6sense's 2025 Buyer Experience Report, based on more than 4,000 buyers, found that 94% of buying groups ranked preferred vendors before first contact with a seller (6sense via Business Wire).
Buying committees also share privately by nature. A champion forwards your pricing page to the finance lead, drops a case study into an internal channel and asks a friend at another firm what they use. Every one of those moments shapes the deal, and none of them leaves a tracking trail.
For firms between $1M and $20M in revenue, this matters even more. Referrals, founder-led content and reputation in a niche are usually the strongest sources of pipeline, and they're exactly the sources dark social hides.
Three methods to measure dark social
No single method catches everything. Used together, they give you a reliable picture.
Method | What it catches | What it misses | Effort |
|---|---|---|---|
1. Self-reported source | Podcasts, peers, communities, AI assistants, events | Exact timing; buyers who forget or simplify | Low |
2. Tagged links for referral paths | Partner, referral programme, podcast, newsletter and post links that do get clicked | Anything shared by word of mouth or copied without the tag | Low to medium |
3. Trend signals | Lifts in direct traffic and branded search after activity you can date | Individual deals; it shows patterns, not causes | Medium |
Method 1: Ask the buyer
Self-reported attribution is the only method that sees dark social directly. Add a required, open-text "how did you hear about us?" field to demo and contact forms. It's the core of the survey method, covered in detail below.
Method 2: Tag every referral path you control
Some dark social can be made visible. Give every path you control its own tagged link using consistent UTM parameters:
Referral and partner programmes: a unique link per partner, so each referral arrives with its source
Podcasts and webinars: a short, memorable page such as yoursite.com/podcast that redirects with UTMs attached
Newsletters and your own emails: tag every link, so forwarded copies still carry a source when clicked
Founder and team LinkedIn posts: tag links in posts and comments
Sales assets: tag the links in decks and one-pagers that champions forward internally
AI assistants are starting to help here too. OpenAI says "ChatGPT automatically includes the UTM parameter utm_source=chatgpt.com in referral URLs" (OpenAI Help Center). That catches clicks from cited links, though not buyers who read the answer and search your name later.
Method 3: Watch the trend signals
Dark social leaves footprints in aggregate even when individual visits are invisible. Track direct traffic to key pages, branded search impressions in Search Console and new inbound demo requests each week. Then mark the dates of activity you can't tag: a podcast release, a conference talk, a founder post that spread or a mention in a community newsletter. A clear lift in direct and branded traffic in the following week or two is a strong sign that activity worked.
// The Rule
You can't track dark social, but you can ask about it, tag around it and watch its footprints. Do all three before you cut a channel that "isn't working".
Setting up the survey method
The self-reported question only works if it's asked well and the answers are used. Here's the setup we use with clients:
Ask in three places: on the demo or contact form, again on the first sales call ("what made you reach out now?") and once more at onboarding for closed-won customers
Keep the form question open text: dropdowns push people toward "Google search" because it's the first sensible option
Ask for specifics: "how did you first hear about us?" gets better answers than "source"
Store it in the CRM: a field on the contact and the deal, next to the tracked original source, so the two can be compared
Code the answers monthly: group free text into a short list such as referral, podcast, LinkedIn, community, AI assistant, event, search and other
Sales call notes and win-loss analysis add depth. When buyers explain why they chose you, they often name the peer, post or episode that started it.
Reading the signals each month
Put the three methods side by side in a short monthly review:
Self-reported vs tracked: for deals created this month, how many show "direct" or "branded search" in the CRM but name a specific channel in the self-reported field? That gap is your measured dark social.
Tagged referral paths: which partner links, podcast pages and post links produced visits, leads and deals?
Trend lifts: did direct and branded search rise after dated activity?
Pipeline by self-reported source: judge channels on qualified pipeline and revenue over rolling quarters, not single months
If you run HubSpot, our guide to HubSpot attribution shows how to build the self-reported vs tracked report. Our guide to first-party data attribution covers the CRM setup underneath it.
Mistakes to avoid
Treating all direct traffic as dark social. Some of it really is people typing your URL or returning from a bookmark. Use self-reported answers to estimate the share, not assumptions.
Over-tagging. Long UTM links in a LinkedIn post or a podcast description look spammy and get shared less. Use short redirect pages instead.
Asking once and never reading the answers. An uncoded free-text field is just noise in the CRM.
Giving self-reported answers final authority. Buyers simplify. Use the answers to spot channels tracking misses, then confirm with trends and, for big decisions, incrementality testing.
Chasing perfect visibility. Some sharing will always stay private. The aim is to make better budget calls, not to see everything.
How mrktrs measures dark social
We build dark social measurement into every client's measurement stack: a self-reported source field on every form and deal, short tagged pages for podcasts, partners and referral programmes, a weekly view of direct and branded search against dated activity and a monthly review that compares what buyers say with what tracking shows. It sits alongside the rest of our attribution work, so founder-led content, referrals and AI visibility get judged on the pipeline they actually create.
Not sure how much of your pipeline starts in conversations you can't see? Book a 30-minute call and we'll show you how to measure it.
Frequently asked questions
What does dark social mean?
It refers to content sharing and recommendations that pass through private or untrackable routes, such as messaging apps, email, closed communities and spoken conversations, so the resulting visits carry no referral data.
Is dark social the same as direct traffic?
No. Direct traffic is a reporting category, while dark social is a behaviour. A portion of direct visits comes from privately shared links, but direct also includes typed URLs, bookmarks and returning visitors.
Can dark social be tracked?
Not completely. Companies can surface part of it through tagged links on the routes they own, buyer surveys and lifts in branded search, but private conversations themselves remain invisible to analytics tools.
Why is dark social important for B2B companies?
B2B purchases involve several decision-makers who rely heavily on peer advice and internal sharing. Ignoring that influence leads teams to undervalue referrals, communities and thought leadership in budget decisions.
Do AI assistants count as dark social?
Partly. When a buyer clicks a cited link, some assistants pass a tracking tag. When the buyer simply reads the answer and later searches the brand, the influence is hidden, which behaves like dark social.