The dark funnel is the part of the B2B buying process you cannot track: a podcast mention, a Slack thread in a private community, a peer recommendation over lunch, a LinkedIn post someone screenshotted into an internal chat. None of it shows up in your attribution tool. Then a prospect fills in a demo form, your CRM credits "Direct" or "Organic search", and the real story is lost.
Most writing about the dark funnel stops at "it exists, so invest in brand". That is true but not very useful on a Tuesday. This post takes a narrower view: the dark funnel is dark everywhere except one place, your website, where it surfaces before the buyer ever talks to you. Here is how to see it there and what to do about it.
How Big the Dark Funnel Actually Is
Three sources point in the same direction.
- 6sense's 2024 Buyer Experience Report found that B2B buyers are nearly 70% through their purchasing process before they engage sellers, and that 81% already have a preferred vendor at the time of first contact.
- Gartner's research on the B2B buying journey found that buyers spend only about 17% of their purchase time meeting with potential suppliers, split across every vendor they are considering.
- LinkedIn's B2B Institute describes the 95-5 rule: at any given time, about 95% of your potential buyers are not in-market.
Put together: most of the decision happens before you hear about it, and the influences that shaped it happened long before that, while the buyer was not even shopping.
What Most Teams Get Wrong About It
The common reaction is to try to measure the dark funnel with better attribution: more touchpoints, multi-touch models, longer lookback windows. That does not work, because the influential touches never generate a trackable event in the first place. No model can assign credit to a conversation it never saw.
The opposite reaction, "attribution is dead, just trust brand", goes too far. It leaves marketing with no feedback loop at all.
We think the useful position sits between those. Stop trying to attribute the dark funnel. Start detecting its output. Dark-funnel influence produces observable side effects: companies that show up on your site with no campaign touch, people who arrive by typing your URL, buyers who go straight to pricing on their first visit. Those effects are measurable, and they are actionable even when the cause is not.
Where the Dark Funnel Surfaces on Your Website
Your site is the one place almost every buyer passes through during the hidden part of the journey. Our 2026 B2B personalization research found that 62% of practitioners said the vendor website was the single most important channel during their own most recent B2B purchase. Here are the signals worth instrumenting.
1. Identified companies with no campaign history
When company-level identification resolves a visitor to an account that has never clicked an ad, opened an email, or been contacted, something outside your tracked channels sent them. Keep a running list of these "unsourced" accounts. Its size is the most direct measure of dark-funnel demand you will get.
2. Direct and branded-search arrivals on deep pages
Someone who types your domain and lands on the homepage might be anyone. Someone who arrives directly on /pricing or /integrations/salesforce was sent a link, usually in a private message. Track direct arrivals by landing page, not in aggregate.
3. First-visit behavior that looks like a fifth visit
A new visitor who goes homepage, pricing, security, comparison page in one session did their research elsewhere. Flag sessions where the first visit already contains late-stage pages.
4. Several people from one company arriving close together
When three people from the same company visit within a few days, with no campaign touch, someone shared your site internally. That is the dark funnel turning into a buying group.
5. Self-reported attribution
Add a free-text "How did you hear about us?" field to demo forms. Not a dropdown, which forces people into your channel list. Free-text answers like "a friend at another company", "saw it in a community thread", or a podcast name are the most direct view of dark-funnel sources you can get. Review them monthly and tag them by hand.
Example: One Account's Hidden Journey
Here is a composite of a pattern we see often, with the parts your tools record and the parts they miss.
| When | What actually happened | What your stack recorded |
|---|---|---|
| Month 1 | A RevOps manager hears your product mentioned on a podcast | Nothing |
| Month 2 | She asks about it in a private RevOps Slack community; two peers recommend it | Nothing |
| Month 3, day 1 | She types your URL and reads pricing and the Salesforce integration page | One direct session, anonymous |
| Month 3, day 2 | She shares the pricing link with her VP and a sales ops colleague in an internal chat | Two more direct sessions, landing on /pricing |
| Month 3, day 9 | The VP clicks a retargeting ad and books a demo | Demo request, attributed to paid social |
The attribution report credits paid social. The real drivers were a podcast and a community thread. But notice what was visible: three people from one company, arriving directly on deep pages, a week before the demo request. With company identification in place, that account would have surfaced on day 2, not day 9, and the free-text source field would have recovered the podcast and the community.
What the Dark Funnel Is Not
The term gets stretched, so it is worth being precise. The dark funnel is not:
- Anonymous website traffic. Visits to your site are trackable. They are just unidentified until you resolve them. That is a data gap, not dark funnel.
- Bad UTM hygiene. If your own campaigns show up as "direct" because links were not tagged, fix the tagging. Do not call it dark.
- A reason to stop measuring. Some channels are unmeasurable. Most of your funnel still is.
Keeping the definition tight stops "dark funnel" from becoming the explanation for every reporting problem.
How to Act on Dark-Funnel Signals
Detection is only useful if it changes what happens next. Three responses, in order of effort:
Make the first visit count
A buyer arriving from a private recommendation already has a shortlist in mind. Your site has one or two visits to confirm you belong on it. Show proof from their industry and size band immediately, not after they click through to the case studies page. Markettailor's visitor identification resolves the company on the first pageview, which is what makes this possible for unsourced accounts that never touched a campaign.
Tell sales which accounts surfaced
Send the account owner a weekly list of unsourced, high-fit accounts that showed late-stage behavior. Include the pages viewed, not a score. A rep who sees "three visitors, read pricing and the HubSpot integration page" knows how to open the conversation. Our post on aligning sales and marketing with visitor data covers the handoff format that works.
Feed the channels that create it
Self-reported answers tell you which communities, podcasts, and peer networks keep coming up. That is where to put more effort: show up in those communities, appear on those podcasts, make your best content easy to share as a link. You will not be able to attribute it precisely. You will see the unsourced-account count move.
Measuring the Dark Funnel Without Pretending to Attribute It
Replace the attribution question ("which channel gets credit?") with trend metrics you can track monthly:
- Unsourced ICP accounts per month: identified, high-fit companies with no tracked campaign touch.
- Direct deep-page arrivals: direct sessions landing on pricing, integration, or comparison pages.
- Self-reported source mix: share of demo requests citing word of mouth, community, podcast, or social.
- Branded search volume: from Google Search Console, as a proxy for people hearing your name elsewhere.
- Pipeline from unsourced accounts: opportunities created from accounts that first appeared without a campaign touch.
Report them as a trend over at least two quarters. Month-to-month swings are mostly noise, especially for smaller sites where a single conference or viral post can double one month's unsourced count.
If your brand and community work is working, these numbers rise together. If they flatten, the activity is not reaching buyers. That is a feedback loop, even without a credit model. Our post on marketing attribution for personalized websites covers how to keep these trend metrics alongside your existing attribution reporting rather than replacing it.
A 30-Day Starting Plan
- Week 1: add a free-text "How did you hear about us?" field to every demo and contact form.
- Week 2: set up company-level identification and build the unsourced-accounts list.
- Week 3: create a segment for first visits that include late-stage pages, and show it industry-specific proof.
- Week 4: send sales the first weekly list of unsourced high-fit accounts, with pages viewed.
The dark funnel will never be fully visible. It does not need to be. You need to notice when it delivers someone to your site, and treat that visit like the late-stage moment it usually is. To see which unsourced accounts are on your site right now, book a walkthrough from our pricing page.