In this post:
The dark funnel is the buying activity that shapes a purchase but never shows up in your analytics or your CRM: the peer recommendations, private messages, podcasts, community threads and internal debates that happen before a buyer fills in a form or answers an email.
You notice it when a lead books a demo already knowing your pricing model, the name of a client you worked with and the exact problem you're good at, while the source field in the CRM says "Direct". Somebody convinced that person. Nothing you can see in a report did.
The gap matters because budgets get set from the visible part. When reports can only credit what they track, the channels that do the early persuading look worthless on paper, and they're usually the first ones cut.
TL;DR
Most of the persuading in a B2B deal happens before the first conversation. 6sense's 2025 survey of nearly 4,000 buyers put first contact with a vendor at about 61% of the way through the buying process, and 77% of buyers said their first vendor conversation was with the one that won.
The dark funnel holds private messages and group chats (dark social), Slack and community threads, podcasts, newsletters, events, review sites read without logging in, questions typed into AI assistants and the buying committee's own internal talk. Bought intent data picks up only a small part of it.
Attribution software can only split credit between touches it recorded. A deal that started on a podcast shows up as direct traffic or branded search, so a rising branded search line is usually a symptom of demand created somewhere else, not a win for SEO.
You can't track it, but you can estimate it: a required free-text "How did you hear about us?" field, the same question on the first call, keyword searches across call recordings, and on/off tests on one channel for 8 to 12 weeks.
Keep funding the channels that create demand even when they can't show a per-deal return, judge them on trend lines over two or three quarters, and expect prospects who met your ideas in the dark to answer outbound more warmly.
The longer and more expensive the sale, the bigger the dark funnel. A committee purchase with a six-month cycle needs a plan for it from day one, while a cheap product bought in one sitting can lean on its tracked channels.
One deal, as the buyer remembers it and as the CRM recorded it
Take a Head of RevOps at a 200-person software company. In March she hears a founder interviewed on a podcast during her commute and remembers one line about lead routing. In April someone in a RevOps Slack community asks for help cleaning up HubSpot, and two people recommend the same firm. In May she reads three of the founder's posts on LinkedIn, screenshots one and sends it to her VP of Sales in a direct message with the words "this is our problem".
In June she types the company's name into Google, clicks the first result and books a call.
The CRM shows one record. Original source: organic search. First touch: June. Four months of persuasion, and none of it left a mark, because none of it could. A podcast doesn't set a cookie, a Slack recommendation doesn't carry a UTM tag, and a screenshot in a DM isn't a click.
This isn't a broken tracking setup. The tools did their job. The dark funnel is just a description of where most buying decisions get made, which is anywhere the seller isn't in the room.
If that sounds abstract, think about the last piece of software you bought for work. You probably asked a peer, read a thread, watched a video someone sent you and talked it over with a colleague before you ever visited the vendor's pricing page. The vendor saw the pricing page visit. That's all they saw.
Why buyers do their homework out of sight
B2B buyers would rather not talk to sales until they've made up their minds, and they mostly get their way. In the 6sense 2025 Buyer Experience Report, the point of first contact with a vendor sat at about 61% of the way through the buying process, down from about 69% a year earlier. The same survey found that 77% of buyers held their first vendor conversation with the company they ended up buying from. 6sense sells intent data, so it has a stake in that story, but the direction matches what buyers tell us on discovery calls: by the time they talk to you, there's a favourite.
Trust plays a part too. Buyers believe a peer who has used a tool over a vendor who sells it, so they take their real questions to places vendors can't read, like private communities, group chats and conversations with former colleagues.
And the process itself doesn't run in a straight line. Google's 2020 research on the messy middle described shoppers looping between exploring options and narrowing them down, many times over, before buying. That study looked at consumer purchases, but anyone who has watched a B2B deal stall for a month and then come back with three new stakeholders will recognise the loop. Most of those loops happen off your channels.
What's inside a dark funnel
"Untracked traffic" undersells it. The dark funnel is made of specific places, and naming them helps you decide where to show up.
Private shares. Links, screenshots and voice notes passed around in DMs, WhatsApp groups and email forwards. This layer has its own name, dark social.
Communities. Slack groups, Reddit threads, private LinkedIn groups and founder circles where someone asks "who do you use for this?" and gets five frank answers.
Audio and inbox content. Podcasts and newsletters, which shape opinions for months and almost never produce a click.
Rooms. Conferences, dinners, meetups and the tool someone used at their last job.
Review and comparison sites such as G2, read without logging in.
AI assistants. A buyer asks ChatGPT for a shortlist of vendors and gets an answer built from whatever has been written about you. Nothing reaches your analytics.
The buying committee's own talk. Your champion forwards a case study to finance, and the real argument happens in a meeting you'll never attend. The bigger the buying committee, the more of this there is.
Bought intent data catches a sliver of this, mostly topic research on publisher networks and activity on review sites. Tools for website visitor de-anonymization can name some of the companies browsing your site. Neither tells you that the browsing started because a peer said your name in a Slack thread. Yet all of it shapes buying intent, usually more than the channels you can measure.
Dark funnel vs attribution: where one stops and the other starts
These two get blurred, so here's the split. Multi-touch attribution takes the touches your systems recorded and decides how to share the credit between them. Its models argue about the weights, but they all work from the same list. The dark funnel is everything that never made the list.
That's why a smarter attribution model doesn't solve it. Go back to the Head of RevOps. Last-touch, linear and W-shaped attribution all hand 100% of that deal to branded search, because branded search is the only touch on record. Splitting nothing more fairly is still nothing.
The demand surfaces in two buckets, branded search and direct traffic. Google Analytics files traffic that, in its own words, doesn't have a clear referral source under "(direct) / (none)", and Google lists missing UTM tags, redirects, link shorteners and ad blockers among the reasons a visit lands there. A link copied out of a Slack thread or typed in from memory carries no tags at all.
So when branded search and direct demo requests climb and nobody can explain why, the honest read isn't "our SEO is working". It's "something upstream is creating demand, and we should find out what". Teams that miss this move budget towards the last visible step and away from the work that created the demand, and a blind spot that looks like a confident dashboard does more damage than one everybody knows about.
How to measure a dark funnel without pretending to track it
You won't get a per-touch ledger here, and you don't need one. What you need is a rough, repeatable read on which hidden channels are pulling their weight. These are the methods we set up, roughly in order of effort.
Ask on the form. Add a required free-text field to your demo form: "How did you hear about us?" Free text beats a dropdown, because a dropdown nudges people towards "Google search" when the real answer is "my old boss uses you". Store the raw answer in a CRM property and tag it into ten or so buckets once a week. This is self-reported attribution, and it's the cheapest view into the dark you'll get.
Ask again on the first call. Form answers are often lazy. A rep asking "what made you reach out now?" gets a fuller story, and logging it in a second property keeps it separate from the form answer.
Search your call recordings. If you record calls with something like Gong, search the transcripts for words like "podcast", "recommended", "newsletter" and "a friend". Buyers name sources out loud that they'd never type into a form.
Compare the two sources. Put the self-reported source next to the tracked source for every new opportunity. The share of deals where they disagree is the closest thing you'll get to a size for your dark funnel.
Run on/off tests. Pause a channel, or double it, in one region or segment for 8 to 12 weeks and watch demo requests and branded search there against the rest. It's crude, and it will tell you more than any model.
On top of those, keep an eye on trend lines: branded search volume in Search Console, direct demo requests and self-reported mentions, plotted against the dates you started a podcast, joined a community or changed your posting rhythm. None of these numbers is exact. Together they're far closer to the truth than a last-touch report.
Should you fund channels you can't prove?
Yes, with rules. The channels that live in the dark funnel (community, content, podcasts, events and a founder who posts with a real point of view) are often your biggest source of demand generation. They create demand, and paid search and outbound capture it later, which is the distinction behind demand capture versus demand creation.
The timing is what scares finance. LinkedIn's B2B Institute says its research shows 95% of your potential buyers aren't ready to buy today. Content that reaches them now pays off when they come into market, which may be a year away. Judging that work on one quarter's pipeline kills it right before it starts to compound.
The rules we'd put around it are simple. Give each dark channel a budget cap and a review date two or three quarters out. Agree in advance what has to move by then, such as self-reported mentions of the podcast, branded search, or inbound from the target segment. If none of it moves, stop. That's a fair test, and it's a lot fairer than cutting a channel because it never gets the last click.
Day to day, think of this work as deposits rather than campaigns. Answer questions in the communities your buyers trust. Get your founder onto the podcasts they listen to. Publish thought leadership with an opinion sharp enough that somebody screenshots it into a DM. Some of it comes back as demand you'll never trace, and that's the deal you're making.
Outbound to people who already know your name
The dark funnel also changes who's on the other end of a cold email. A prospect who heard your founder on a podcast or saw your posts in their feed for months isn't really cold, even if the CRM has no record of them. They're a warm contact with a cold label, and the gap between warm and cold leads is smaller than your data says.
That's why content-led outbound tends to beat a generic sequence. Instead of explaining who you are, the first message can pick up an idea the prospect has probably already met: the argument from last week's post, the framework from the podcast. Founder content pays twice this way, once by creating demand and again by making every outbound message land on a familiar name.
How big your dark funnel is depends on what you sell
Every company has one, but the size varies a lot. The more considered the purchase, the more of it happens out of sight.
If you sell something expensive that a committee researches for months, plan around a large dark funnel from day one. Long cycles, high prices and lots of peer input all push more of the decision into private channels, and your attribution will undercount your best channels badly.
If you sell something cheap that one person buys in a single session, there's less private research to miss. The tracked click is closer to the real start, and leaning on tracked channels won't mislead you as much.
Wherever you sit on that scale, the same four traps come up. Trusting last-touch reports and cutting whatever looks unattributable. Buying another tracking tool in the hope of seeing what can't be seen, when no pixel will ever record a dinner conversation. Killing slow channels after a single quarter. And dismissing the messy "How did you hear about us?" answers because they don't fit a dashboard, when they're often the most accurate source data you own.
The bottom line
The dark funnel isn't a reporting bug you can fix with better software. It's the normal way B2B buyers make up their minds, and it means your reports describe the end of most decisions rather than the start.
Measuring it is about judgment more than precision. Put the buyer's own account next to the software's, read your call notes, test channels by switching them on and off, and fund the work that creates demand on a timeline that gives it a chance to show up.
Teams that get this right stop arguing about which last click deserves the credit and start asking where their buyers spend their time before they ever visit the website. That question leads to better budgets.
If you'd like help setting up the source field, the call tagging and a report that puts self-reported and tracked sources side by side, Nebor builds it in the CRM you already run and hands it over with notes on how to read it.
Frequently asked questions
What is the difference between the dark funnel and dark social?
Dark social is one part of the dark funnel. It covers links and content shared privately, in direct messages, group chats, email forwards and messaging apps, where analytics tools can't see the referral. The dark funnel is wider. It also includes podcasts, events, community threads, review sites, AI assistant answers and the internal debate inside the buying committee. Every dark social share sits in the dark funnel, but plenty of the dark funnel never involves a shared link.
Is the dark funnel the same thing as the messy middle?
No, although they overlap. The messy middle comes from Google research on consumer purchases and describes how people loop between exploring options and narrowing them down before they buy. It's about the shape of the decision. The dark funnel is about visibility: the part of that decision a seller can't observe. Much of the messy middle happens in the dark, but some of it, like repeat visits to your pricing page, is perfectly trackable.
Can intent data or website visitor identification reveal the dark funnel?
Partly. Intent data shows that a company is researching a topic across publisher networks or review sites, and visitor identification can name some of the companies browsing your website. Both are useful for timing outreach. Neither tells you why that research started, who recommended you, or what was said in a private community. Treat them as a few lit windows in a dark building rather than a floor plan.
Are AI assistants like ChatGPT part of the dark funnel?
Yes, and a growing part. When a buyer asks an AI assistant for a shortlist of vendors or a comparison of approaches, the answer draws on what has been published about you, and the conversation leaves no trace in your analytics. If the buyer later visits your site, it usually shows up as direct or branded search. Asking new leads where they first heard of you is currently the most practical way to spot it.
Will a "How did you hear about us?" field hurt form conversion?
One extra field rarely makes a noticeable difference on a demo request form, where the visitor already intends to talk to you. On a content download it can cost more, so many teams only add it to high-intent forms. Keep it short, make it free text rather than a long dropdown, and test it for a few weeks if you're worried. The insight it gives is usually worth far more than a small dip.
How long does it take for dark funnel investment to show up in pipeline?
Expect months, not weeks. A podcast appearance or a strong post may reach buyers who won't be in market for a year, and community presence builds trust slowly. Early signs usually come first in self-reported answers that name the channel, then in branded search and direct demo requests. If none of those signals move after two or three quarters of steady effort, that's a fair reason to rethink the channel.
How should sales handle a lead that came through the dark funnel?
Start by finding out what they already know. A lead who heard your founder on a podcast or got a peer recommendation often arrives with a shortlist and a view on the problem, so a standard discovery script can feel slow. Ask what made them reach out now and who mentioned you, log the answer in the CRM, and spend the call on their specific situation rather than your company introduction.
Is it worth buying a tool to track the dark funnel?
Not if the aim is to see everything. Private conversations, podcasts and internal meetings can't be tracked by any tool, and vendors that promise otherwise usually mean intent data or visitor identification under a new name. Those can still help with timing and account selection. For understanding where demand starts, a free-text source field, call notes and channel tests will tell you more for far less money.
Is the dark funnel only a B2B problem?
No, consumers ask friends, read forums and watch reviews long before they buy too. It hurts B2B more because the purchases are larger, involve several people and take months, so more of the decision happens in private conversations and the gap between what's tracked and what happened grows. A low-cost consumer product bought in one sitting has a much smaller dark funnel than enterprise software.
Share this post
Intent
Marketing
Glossary
Related glossary items
Understanding the core GTM concepts
Ready to build your pipeline
