The History of the Sales Funnel

By Victor Hoang · Updated Oct 24, 2026

The problem in the early 1920s was how to move a buyer from a broad, general interest to a single yes. A bond buyer faced a mass of facts about securities. He could only act on one idea at a time.

The sales funnel is usually traced to William W. Townsend's 1924 book Bond Salesmanship, which a 2012 ClickZ column credits as the first work to picture a sale as a funnel. It spent most of the following century as a drawing on a whiteboard. It became a managed system only when software and demand-generation teams started counting people at each stage in the 1990s and 2000s. The one-line throughline: a simple picture of narrowing became a measurement tool, then a target of criticism, and it is still how most revenue teams talk.

The funnel image answered that problem as a picture of narrowing attention. It did not answer it with data. For that, sales had to wait for the process thinkers of the 1980s and the databases that followed. For the wider story, see the history of sales.

Timeline at a Glance

The Staircase Idea (Before 1924)

Buyers climb steps, but nobody draws a funnel

Before the funnel there was the staircase. The ClickZ account says people had long viewed a sale as a series of steps from awareness to decision, but no one had pictured it as a funnel. The best-known version of that staircase is the attention-interest-desire-action formula from advertising and sales training, whose authorship is debated. It has its own article: see the history of the AIDA model.

The practical pressure came from the sales floor. When John H. Patterson bought control of NCR in 1884, his early direct mail to 5,000 prospects had produced almost no response, which is why he built a trained field force instead. That story is told in the NCR sales training history. By 1900 the company was setting district quotas, covered in the history of sales quotas and territories. Quotas are a count of outcomes. A funnel is a count of people on the way to the outcome. The first was in place; the second was not.

Keep precursor and concept apart. Merchants in the ancient world kept records of goods, debts and agents, and that earlier history is covered in the quotas article. Selling also long involved sorting casual browsers from serious buyers: think of a catalog or a doorstep call, as in the history of catalog selling and the history of door-to-door sales. But sorting by instinct is not a funnel. The funnel needs two things that arrived later: stages with names, and a way to count how many people sit in each.

For teams today: the first job is not to draw the funnel. It is to agree on the steps a buyer takes in your market. If two reps would describe those steps differently, you have a staircase argument, not a funnel.

Townsend's Funnel (1924-1970s)

A bond textbook gives selling its shape

The turning point is a book. In 1924, according to ClickZ, William Townsend wrote Bond Salesmanship, a 468-page guide for aspiring Wall Street brokers. The column quotes his description of a sale as forcing "a broad and general concept of facts through a funnel" that ends in a single favourable consideration. Read that sentence as a teaching device for a bond salesman. The funnel is not a pool of customers on a dashboard. It is the narrowing of what a buyer thinks about, from a general picture to one fact that leads to a decision.

Two cautions belong here. The claim that Townsend was first rests on trade-press accounts of the book, and the book itself has largely dropped out of view. The wider context also matters. The same column says people already saw a sale as a staged process. What Townsend contributed was the image, and images last. For decades the funnel stayed a teaching image, a way to explain narrowing attention and not a number on a manager's report.

The funnel of this era is a picture of attention. It describes what a buyer's mind does, and it says nothing about a manager's forecast. The new problem was obvious in hindsight: a drawing cannot tell you where your own deals are.

For teams today: a funnel diagram that has no numbers on it is a Townsend funnel. It explains, but it cannot manage. Put a count under every stage before you call it a tool.

The Process Years (1980s-1990s)

Selling becomes a stage-by-stage system

The next turn came from a change in how selling was taught. In 1985, Robert Miller and Stephen Heiman published Strategic Selling, which its publisher describes as presenting selling as a joint venture and emphasizing "process". The book belongs to the same period as SPIN, covered in the history of SPIN Selling, and to the rise of the full sales process, told in the history of solution selling. Our reading, not a documented claim from those sources: once selling is seen as a process with stages, a funnel can be a manager's instrument and not just a drawing.

Software finished the job. Salesforce says its journey began in February 1999 with a vision of a world-class internet company for sales force automation, and that it incorporated on 8 March 1999. From then on a deal stage was a field in a shared database. A manager could see how many opportunities sat at each stage, and the pipeline, in the sense of a counted funnel of live deals, became something a team could read on a screen.

The new problem was a split. Marketing owned the top of the funnel and sales owned the bottom, and each team counted differently. A lead that marketing called ready was often one that sales called noise. The next era tried to fix that handoff.

For teams today: your CRM stages are a funnel only if each one has an entry rule. Write the rule down. The sales funnel stages guide shows what each stage should mean.

The Waterfall Years (2002-2012)

Marketing and sales agree on a handoff

The pivot is a shared vocabulary. A MarketingProfs article from 2012 notes that since 2002 many marketers had used SiriusDecisions' processes and its terminology of marketing-qualified lead (MQL) and sales-qualified lead (SQL) to align sales and marketing activity and track results. A SiriusDecisions research director, Terry Flaherty, is described in a CustomerThink interview as the acknowledged godfather of the SiriusDecisions Demand Waterfall.

The waterfall was a funnel with handoffs. It cut the pipeline into stages and gave each handoff a name and an owner. MQL meant marketing had judged a lead worth sales attention. SQL meant sales had agreed. The same words are now in almost every CRM, and the agreed funnel model explains how teams use them to settle disputes about lead quality. The lifecycle view of the same stages is in revenue funnel stages, and the SaaS version is in the SaaS marketing funnel.

By 2012 SiriusDecisions had released a new version of the waterfall, and MarketingProfs wrote that what was once a tidy "funnel" or a simple "waterfall" was no longer as relevant, suggesting a river fed by many streams instead. The model that made the funnel countable was already being revised to fit a messier reality.

For teams today: MQL and SQL work only when both sides sign the definition. Review the rule once a quarter, because the rule that matches your buyers this year will drift.

The Journey Challenge (2009-2018)

Researchers and vendors argue the funnel is the wrong shape

This era overlaps the last one. While marketers were refining the waterfall, researchers were questioning the picture under it. In the third issue of McKinsey Quarterly for 2009, David Court, Dave Elzinga, Susan Mulder and Ole Jørgen Vetvik argued that the funnel concept fails to capture all the touch points and key buying factors of a market with many channels and well-informed consumers. They examined the purchase decisions of almost 20,000 consumers across five industries and three continents, and proposed a circular journey with four phases: initial consideration, active evaluation, closure and postpurchase.

Note what they did and did not say. The authors wrote that the funnel metaphor does help a good deal for comparing a brand with its competitors at each stage and spotting bottlenecks. Their evidence came from consumer industries such as autos, skin care, insurance, consumer electronics and mobile telecom, not from complex business-to-business sales. So the finding is a warning to test the funnel against your own buyers, not a verdict that every funnel is wrong.

In September 2018 the argument became a slogan. HubSpot's own post on the change says that funnels lose the energy you put into them once you reach the bottom, whereas a flywheel stores it. Its CEO, Brian Halligan, retired the funnel in his keynote at INBOUND 2018, and CMSWire reported him saying that "We filled that darn funnel with friction." HubSpot sells software built around the flywheel, so read this as a company explaining its own positioning. It is still a real turning point, because it shows the customer after the sale as part of the growth model, which the classic funnel leaves out.

The challenge produced a new problem of its own. A loop is harder to forecast than a funnel, and most sales managers still need a number for next quarter.

For teams today: treat the funnel and the journey as two views, not rivals. Use the funnel to count and forecast, and use a journey map to understand why a buyer stalls.

The Agent Age (2020s-present)

Software starts working the top of the funnel

The funnel has survived its critics because every system built on it still needs stages. The change in the 2020s is who does the work at the top. In its State of Sales report, published 3 February 2026, Salesforce surveyed more than 4,000 sales professionals, and said 48% lack the bandwidth for adequate cold outreach despite devoting nearly a full day a week to prospecting. Sellers who use agents expect them to cut prospect research time by 34%. These are the survey's figures, from a vendor that sells agents, and they are expectations, not measured results.

If that holds, the old split between "top of funnel" and "bottom of funnel" teams will blur. The funnel then moves from a picture of buyers to a picture of work: which tasks are done by people, which by software, and where the human judgment still decides. The next stage of the funnel may be defined less by how buyers narrow their choices and more by what a team automates between stages.

For teams today: before you add an agent, write down the stage it serves and the number it should move. The tool is new, but a stage with no entry rule is just as hard to automate as it is to manage.

Key People

Key Facts

• ClickZ credits Bond Salesmanship (1924) with the first picture of a sale as a funnel.
• Salesforce incorporated on 8 March 1999 with a vision built on sales force automation.
• MarketingProfs says MQL and SQL terminology had been in use since 2002.
• McKinsey's 2009 study drew on almost 20,000 consumers in five industries.
• HubSpot announced the flywheel at INBOUND 2018.
• In a 2026 Salesforce survey, 48% of sellers said they lack bandwidth for adequate cold outreach.

The Pattern Behind the History

Our synthesis, not published research. Call it the Picture-Count-Challenge-Automate cycle:

• Picture. Someone draws a simple shape that explains how buyers narrow their choices.
• Count. Process thinkers and software turn the shape into stages with numbers.
• Challenge. Researchers and rivals show buyers don't move in a straight line.
• Automate. Tools take over the repetitive work between stages, and the cycle restarts.

The funnel went through each step once and is entering the second round.

What It Teaches Teams Today

• Name the steps first. A funnel only works when your team agrees on what each stage means.
• Count at every stage. A drawing explains; a number manages.
• Write handoff rules. MQL and SQL succeeded because they put a name on a handoff.
• Test the shape against your buyers. McKinsey's critique came from consumer markets. Your market may still be a funnel.
• Keep the customer after the sale in view. The flywheel's real point is that growth continues after the close.

For the lineage around it, read the evolution of selling, the dated sales timeline and the history of the Challenger Sale.