What the funnel is actually a model of
The marketing funnel is a population model. It describes a set of potential buyers narrowing as they pass through stages, and its real content is the conversion rate between one stage and the next. It is a counting device with a shape.
It gets confused with three other things. It is not a description of how an individual buyer thinks; nobody moves cleanly from awareness to consideration to purchase. It is not a process, though sales stages are usually bolted onto it. And it is not a demand-generation strategy, though budgets are often structured as though "top of funnel" and "bottom of funnel" were categories of work rather than positions in a count.
The distinction matters because the funnel is genuinely useful for one job and misleading for most others. As arithmetic — how many enquiries produce how many opportunities produce how many wins — it earns its place. As a theory of buyer behaviour it is wrong, and Gartner's own research says so.
The standard stages and what each is supposed to represent
Terminology varies by company; the underlying stages do not. A common B2B version:
| Stage | What it represents | Typical evidence |
|---|---|---|
| Awareness | The buyer knows the category and, ideally, your name in it | Branded search, direct traffic, unaided recall |
| Interest | Active information gathering about the problem | Non-branded search, content consumption, newsletter signup |
| Consideration | You are on a list of possible answers | Comparison and pricing pages, demo requests |
| Intent | A specific purchase is being scoped | Meetings booked, requirements documents, RFP invitations |
| Evaluation | Formal assessment against named alternatives | Trials, security review, references requested |
| Purchase | Contract, procurement, signature | Closed-won |
Two things are worth stating plainly. The stage names describe your visibility of the buyer rather than the buyer's mental state, because you are inferring interest from behaviour observable on your own property. And the evidence column is where most implementations fall apart: companies define stages by marketing artefact rather than by anything the buyer has committed to.
What it is genuinely good for
Three uses survive scrutiny.
A shared vocabulary. Marketing, sales, and finance need one set of words for the same objects. When a CFO asks how many opportunities marketing produced last quarter, the funnel supplies definitions that make the answer checkable. This is unglamorous, and it is the biggest single reason the model persists.
A conversion diagnostic. If enquiry-to-meeting is healthy and meeting-to-opportunity collapses, you have located a problem — probably qualification, or a mismatch between what marketing promises and what sales sells. The funnel does not tell you the cause, but it narrows where to look, which is more than most dashboards manage.
Capacity and cost arithmetic. With stage rates and average deal value you can work backwards from a revenue target to a required volume of enquiries, and forwards from a budget to a plausible outcome. Both calculations are rough. Rough beats absent when somebody has to sign a budget.
All three treat the funnel as accounting rather than psychology. Used that way it is a decent tool.
Where the metaphor starts lying to you
Three failures, in order of the damage they do.
Buying is not linear. Gartner describes B2B buying as six jobs rather than stages — problem identification, solution exploration, requirements building, supplier selection, validation, and consensus creation — and states that buying does not play out in any predictable, linear order, with customers looping back and revisiting jobs across a purchase. A buyer can be writing requirements and re-exploring solutions in the same week. A funnel that assumes forward motion records that as a stalled deal.
A committee is not a prospect. The funnel counts one thing moving down a tube. The reality is a group of people at different stages simultaneously, several of whom have never visited your website, one of whom can veto the purchase for reasons no measurement will capture. Stage is a property of the account at best, and the account contains disagreement.
It assumes demand already exists. Everything starts with someone entering the top, which quietly reduces marketing's job to routing existing interest. The larger question — how many companies will even recognise they have this problem this year — sits outside the diagram entirely.
Conversion rates and the sample-size problem
Stage conversion rates are the funnel's main output and, in most B2B businesses, they are too noisy to act on.
The arithmetic is unforgiving. Close 40 deals a year from 600 enquiries and a single quarter gives you roughly 150 enquiries and 10 wins. One large deal slipping into the next quarter moves the headline rate substantially while nothing about the business has changed. Segment that by channel and campaign, as every dashboard invites you to, and you are computing percentages on counts of three.
Practical responses, none of them exciting:
- Report rolling twelve-month rates rather than quarterly ones, and show the underlying counts next to every percentage.
- Use cohorts by entry period instead of period totals, so you are not dividing this quarter's wins by this quarter's enquiries — different populations, separated by your cycle length.
- Set a minimum count below which you decline to publish a rate at all, and say so rather than showing a number.
- Watch time in stage as well as rates. Duration is often the earlier signal and is less sensitive to small counts.
The alternatives, described fairly
The demand unit waterfall. SiriusDecisions, now part of Forrester, replaced its own lead-based waterfall with a model whose unit of measurement is the buying group and its identified demand rather than the individual lead. The improvement is real: it stops an organisation counting people when it sells to committees, and it distinguishes demand you have detected from demand you have created. The cost is complexity and a dependence on account-level data most mid-sized companies cannot assemble cleanly.
The bowtie. Common in revenue-operations circles, this mirrors the funnel after the sale — acquisition on the left, then onboarding, adoption, retention, and expansion widening to the right. Where most lifetime value arrives after the first contract, it corrects a genuine blind spot, because a funnel that ends at closed-won makes renewal somebody else's diagram. It inherits the linearity problem, though, and adds a second set of stage definitions to argue about.
Neither replaces thinking. Both are better than the classic funnel at the specific thing they were built to fix, and no model makes a small number of deals statistically informative.
Using the funnel without believing it
The workable position is to keep the funnel as a reporting convention and refuse to let it govern strategy.
- Define stages by buyer commitment, not marketing artefact. "Agreed to a scoping call" is a stage. "Downloaded a whitepaper" is a page view with a form in front of it.
- Allow backward movement and record it. If your CRM cannot show a deal returning to requirements-building, your data contradicts how buying works. Deals that loop are normal, not sick.
- Track at account level, with contact coverage as a separate measure. How many people in the account have you reached, and from how many functions? That is answerable. Asking which stage a fourteen-person committee occupies is not.
- Fund the top of the diagram as demand creation rather than routing. Decide that share deliberately instead of letting the shape of the model imply it should be small.
- Publish counts with every rate. It is the cheapest available defence against confident conclusions drawn from tiny samples.
The funnel is a filing system. Filing systems are useful, and nobody should mistake one for a description of the world.
Frequently Asked Questions
What are the stages of the B2B marketing funnel?
A common B2B sequence is awareness, interest, consideration, intent, evaluation, and purchase. Some companies collapse the middle into a single consideration stage; others add post-sale stages for retention and expansion. Names vary, and what matters is that each stage has a definition somebody can check.
Define them by what the buyer has committed to rather than by what your marketing produced. A stage defined as a content download tells you about your form, not about the purchase.
Is the marketing funnel still relevant in B2B?
As a counting and reporting device, yes. As a description of how businesses buy, no. Gartner's research describes B2B buying as six jobs revisited in a loop rather than a linear sequence of stages, and the funnel cannot represent that.
Keep it for conversion diagnostics, capacity planning, and shared vocabulary between marketing, sales, and finance. Make strategy decisions from a model that accounts for buying groups, non-linear behaviour, and demand that does not yet exist.
Why are our funnel conversion rates so inconsistent?
Usually because the sample is too small for the arithmetic. A business closing a few dozen deals a year has single-digit counts once a quarter is split by channel, and one deal moving between quarters swings the percentage without anything changing in the business.
Use rolling twelve-month figures, group deals into cohorts by when they entered rather than dividing this quarter's wins by this quarter's enquiries, publish counts alongside every rate, and decline to report a rate at all below a stated minimum number of outcomes.
What is the difference between a marketing funnel and a buyer journey?
A funnel is a supplier-side count: how many of the people you can see moved from one stage to the next. A buyer journey attempts to describe what the buying organisation is doing, including work you cannot observe — internal discussion, budget approval, peer conversations, security review.
They answer different questions. The funnel tells you where throughput is lost. The journey tells you what the buying group needs next, which is what determines what you should produce.