What is a sales funnel?

What a sales funnel is

A sales funnel is a counting tool. Picture everyone who could buy from you at the wide top of it, and everyone who actually pays at the narrow bottom. In between are the steps where people drop out: they leave your site, ignore a follow-up, or choose a competitor. The funnel is a way to count those steps and see where the biggest losses are.

The funnel is a model, not a fact of nature. No buyer experiences their purchase as “I am now in the middle of the funnel”. What the funnel gives you is a common language and a set of numbers you can compare month to month.

It connects directly to what online sales is: the funnel describes the path whichever model you use, whether the end of it is a checkout or a conversation.

AIDA and top, middle and bottom of funnel

The two most common ways to name funnel stages are both practitioner conventions. AIDA stands for attention, interest, desire and action. It is a long-standing advertising formula commonly attributed to the early twentieth-century advertising writer E. St. Elmo Lewis, though accounts of its exact origin differ.

Top of funnel (TOFU), middle of funnel (MOFU) and bottom of funnel (BOFU) are the digital marketing version. TOFU is awareness: the buyer has a problem and is searching. MOFU is consideration: they are comparing solutions and providers. BOFU is decision: they are ready to buy and need reassurance and an easy path. There is no standards body behind these labels, so teams define them differently. Write your own definitions down.

• Attention or TOFU: guides, search results, social posts, answers in AI assistants. The job is to be found and be useful.
• Interest and desire, or MOFU: service pages, comparisons, case descriptions, reviews, pricing guidance. The job is to help the buyer evaluate you.
• Action or BOFU: checkout, quote form, booking page, a phone call. The job is to make the decision easy and safe.
• After the sale: delivery, onboarding and asking for a review. Many funnel diagrams stop at the sale, but repeat and referral revenue starts here.

Why real journeys are messier than a funnel

In July 2020, Google published research by Alistair Rennie and Jonny Protheroe, with The Behavioural Architects, describing what they call the messy middle. Between a trigger (something makes you want a product or service) and a purchase, people move between two modes: exploration, where they widen their options, and evaluation, where they narrow them. They can loop between the two many times.

For a business owner this changes three things. First, a visitor who leaves is not necessarily lost: they may be exploring and will return to evaluate. Second, being present in more places during exploration matters. Third, evaluation is influenced by identifiable biases. Google lists category heuristics, the power of now, social proof, scarcity, authority and the power of free.

Google’s experiment simulated 310,000 purchase scenarios. A fictional challenger could take a share of preference from a favourite when it was easier to evaluate and offered more of these cues. That is evidence from a simulation by Google and its partner, not a guarantee for your market, but the direction is useful: make evaluation easy.

See messy middle in the glossary. The honest summary is that a funnel is still a useful measuring tool, as long as you remember that people skip, repeat and leave steps.

Funnel metrics and drop-off maths

For each step, calculate the step conversion rate: the number who reached this step divided by the number who reached the previous step. The drop-off rate is 100% minus that. The overall conversion rate is the final number divided by the first. See drop-off rate.

Here is an illustrative worked example for a service business. Numbers are made up to show the method, not benchmarks. In one month, 8,000 people visit the site. 3,200 view a service page (40%). 400 start the enquiry form (12.5% of those). 240 submit it (60%). 120 enquiries become booked calls that actually happen (50%). 30 become paying clients (25%).

End to end that is 30 out of 8,000, or 0.375%. The biggest percentage drop looks like the service page to form start (87.5% leave), but percentages mislead. Test what a realistic improvement is worth at each step.

• Raise form completion from 60% to 70%: 280 submissions, 140 calls, 35 clients. That is five more clients.
• Raise service-page-to-form-start from 12.5% to 15%: 480 starts, 288 submissions, 144 calls, 36 clients. That is six more clients.
• Raise call-to-client from 25% to 30%: 36 clients. Six more again, and it needs no extra traffic.
• Doubling traffic with no other change: 60 clients, but traffic costs more than most on-site fixes. Fix leaks before buying more visitors.

How to measure a funnel in practice

Google Analytics 4 has a funnel exploration report. Google’s documentation says it maps the steps users take to complete a task and shows where they succeed or drop off. You choose between an open funnel, where people can enter at any step, and a closed funnel, where they must enter at the first. Users count toward a step only if they complete the steps in order, so define your steps carefully.

Your analytics will not see everything. Phone calls, walk-ins and conversations in a CRM (customer relationship management system, your database of contacts and deals) sit outside the website. Join them: record the source of every enquiry and log what happens to it, so the later steps of the funnel are real numbers rather than guesses.

For shops, Baymard Institute’s aggregated research puts average cart abandonment at 70.22% across 50 studies (updated September 2025). Use that as context, not a target. Your own baseline matters more, and the stated reasons, such as extra costs and forced account creation, tell you where to look.

How to build and improve a funnel, step by step

Start from the purchase and work backward. List the actions that must happen for a sale, name each one in plain words, and put a number against it.

• Write the steps: for example visit, view service page, start form, submit form, call held, proposal sent, client signed.
• Count the last 3 months for each step. If a step is not tracked, set up tracking first.
• Calculate the step conversion for each pair and find the lowest ratio that you can influence.
• Form a reason for the drop: unclear offer, too many form fields, slow reply, bad-fit traffic. Look at recordings, form analytics and the replies you send.
• Change one thing, wait for enough volume to judge, and compare on the same measure.
• Repeat monthly. The funnel is a habit, not a project.

When funnel thinking goes wrong

The common failure is optimising a number that is easy to see instead of the one that pays. Form submissions can rise because the form is easier while the quality falls. Always follow a funnel through to closed deals, and see how to qualify leads so a bigger funnel top is not mistaken for progress.

A second failure is forcing every buyer through the same path. Someone returning to buy a repeat order needs a shortcut, not an introduction. A third is assuming a sample is big enough: with 40 form starts a month, a one-week change tells you almost nothing. Wait, or use more careful comparison.

Finally, a funnel measures what you can see. If the first touch was an assistant answer, a friend or a review site, the funnel starts late. Read the lead generation guide for how to widen the top.