Open any B2B marketing funnel report and you get a tidy shape. Wide at the top. Narrow at the bottom. Arrows pointing one way. Your buyers do not move like that. They loop. A buyer goes quiet, then reappears with a colleague who has never heard of you. Much of that happens on pages you do not own and cannot count. The shape is not wrong because it simplifies. It is wrong because it points in a direction buying does not travel.
Buying loops, and a funnel cannot draw a loop
Gartner’s account of B2B buying is the one most of this industry now works from. It describes six jobs a buying group has to finish. Identify the problem. Explore solutions. Build requirements. Select a supplier. Validate the choice. Get everyone to agree.
Look at the order it implies. There isn’t one. Buyers revisit those jobs instead of passing through them once. A stakeholder who joins late can send the whole group back to requirements. That is commonly reported as looping. It is described as how buying works, not as an exception.
A funnel cannot draw that. The shape only points one way. So when a real buyer loops, your report shows them dropping out, or shows nothing at all. Your team then goes hunting for a leak. There is no leak. There is a person thinking.
Six to 10 people are deciding, and your CRM knows a fraction
The funnel also gets the unit wrong. It counts leads, which means individuals. Complex B2B purchases are rarely made by one person.
Gartner commonly reports six to 10 decision makers on a complex B2B purchase. Each one runs their own research. Your CRM probably has an email address for a fraction of them.
Think about who is missing. The security reviewer who killed the deal never filled in a form. The finance director who revived it read one comparison page. Then she asked a question in a Slack channel you will never see. Neither of them is in your report.
So a stage conversion rate measures the wrong object. Deals move when a group agrees. The group is mostly invisible to you.
You are in the room for about 17% of the purchase
Here is the number that should change how you read every funnel chart you own. B2B buyers spend about 17% of the total purchase time meeting with potential suppliers. That is Gartner’s figure, commonly reported. And 17% is the share for all suppliers combined, not for you.
The bigger blocks sit elsewhere. Around 27% of buying time goes to independent research online. That is the single largest slice. Another 18% goes to independent research offline. Add those two and roughly 45% of the process is research you are not present for and cannot instrument.
Data table
| Activity | Share of total purchase time |
|---|---|
| Independent research online | 27% |
| Independent research offline | 18% |
| Meeting with potential suppliers | 17% |
Your funnel report is built out of the moments you were present for. It is not lying to you. It describes a minority of the process with total confidence. That is worse than describing nothing, because confident and partial is what moves budget.
Now split that share across your shortlist
Now split it. Several suppliers are usually in the running. Commonly reported figures put any single one of them at roughly 5% to 6% of the buyer’s total time.
Data table
| Who the time is spent with | Share of total purchase time |
|---|---|
| All potential suppliers combined | 17% |
| Any one supplier, at best | 6% |
Read that from your own chair. Every meeting you will get, plus the demos, the follow-up calls, and every other moment a human from your company is actually in the room: all of it adds up to about one twentieth of the decision.
The rest is your content, your reviews, your pricing page, a competitor’s comparison post, and a conversation in a private community.
A Gartner sales survey from 2025 puts it more bluntly. The result, commonly reported, is that 61% of B2B buyers said they would prefer a rep-free buying experience. That is a stated preference, not an accident of your funnel.
The dark funnel is demand with no receipt
Data table
| Measure | Value |
|---|---|
| Decision makers, top of the range | 10 |
| Of buying time you are not in | 45% |
The pattern is familiar. Someone hears you on a podcast. They see a customer post about you. They read a few articles and sit on it. Then they type your brand name into Google and convert. Your report thanks organic search.
People call this the dark funnel. The name is dramatic for what it is. It is demand generation with no receipt. The work happened, and the demand is real. Your model just hands the credit to whichever touchpoint had the manners to be last.
The damage here is not philosophical. It lands in the budget. Channels that create demand look expensive, because they get credited with almost nothing. Channels that harvest existing demand look brilliant, because they get credited with everything. Cut the first to fund the second. Before long the second stops working too, because nobody is making new buyers.
What to measure when attribution cannot see the journey
You do not fix this with a better attribution model. You fix it by measuring things that survive not knowing.
Ask buyers directly. A required field on the demo form, “how did you first hear about us”, collects what no tracker can reconstruct. The answers are self-reported and imperfect. They still beat a model that was never fed the podcast.
Watch branded search as a demand signal, not a vanity one. Volume rises when people you cannot track decide to look you up. Then test whether pipeline moves at account level when a demand channel moves, allowing for a lag. That is a fairer question than anything a last-click report can answer.
Judge the whole thing the way a board would. There is a fuller version of that argument in the B2B marketing KPIs worth reporting. The short form: pipeline generation and revenue are the honest scoreboard. Neither one asks you to know which touch did the work.
| What the funnel reports | What it cannot see | What to measure instead |
|---|---|---|
| Lead stage conversion rates | The rest of a buying group Gartner puts at six to 10 | Account level engagement and pipeline created per account |
| Last touch channel credit | The roughly 45% of buying time spent in research you are not present for | Self reported source on the form, plus branded search as a demand signal |
| Linear stage progression | Buyers looping back to requirements when a stakeholder joins | Time in stage and re-entry rate, read as normal rather than as failure |
None of this means you delete the stages. They are a useful shared language between marketing and sales. The handoff between the two still has to be built on purpose, which we covered in sales and marketing alignment. Keep the vocabulary. Drop the arithmetic that pretends the vocabulary is a measurement.
The practical version is short. Build for the part of the process you are not in the room for. Run demand generation as though the buyer will decide before they ever speak to you, because on the numbers above that is the safer assumption.
FAQ: the B2B marketing funnel
Is the B2B marketing funnel dead?
The stages are still useful as shared vocabulary between marketing and sales. The geometry is what fails. A funnel encodes one direction, and B2B buying loops back on itself as new stakeholders join and requirements change. Use the stages as labels, not as a model of how buyers move.
Why does the funnel not match what our sales team sees?
Because it counts individuals, and buying is done by a group. Gartner commonly reports six to 10 decision makers on a complex B2B purchase. Your CRM usually holds records for a fraction of them. So stage conversion rates describe a sample, not the buying group.
How much of the B2B buying process happens without a supplier present?
Most of it. Buyers spend about 17% of total purchase time meeting with potential suppliers, commonly reported by Gartner. That share covers every supplier they are considering. Independent research accounts for roughly 27% online and 18% offline.
What is the dark funnel?
It is demand created where your analytics cannot see: podcasts, private communities, word of mouth, review sites, an article someone read months before converting. The demand is real. The buyer then appears in a channel that did not create it, which is how demand generation ends up looking expensive.
How do you measure marketing when attribution cannot see the journey?
Add a self-reported source field to the form and take the answers seriously. Track branded search as a demand signal. Test whether pipeline created at account level moves when a demand channel moves, allowing for a lag. Judge the programme on pipeline and revenue, not on which touch was last.
Should we still use lead scoring and MQLs?
Keep them if they trigger a useful action. Drop them if they only produce a number for a report. The failure mode is treating one person’s score as the state of a buying group. Scoring accounts rather than people gets closer to the thing that actually buys.
COSEOM Team
COSEOM is a multilingual B2B growth agency. Since 2008 the COSEOM team has run international SEO, generative engine optimization, paid media and demand generation programs for B2B software and technology companies, in English, German, Spanish and Portuguese. This account publishes our research, playbooks and product guides. More about COSEOM