What a sales funnel actually is
Every agency deck has the same picture: a trapezoid, wide at the top, narrow at the bottom, divided into four coloured bands. The picture is not the funnel. This guide covers what the four stages actually do, what has to be true at each one, and why the stage most businesses skip is the one that decides whether any of it pays.
The diagram is a picture, not a plan
A funnel is not a shape your customers travel through. It is a sequence of decisions you make on their behalf: what you show them, what you ask them for, what you do while they think about it, and what happens the moment they say yes. Draw it however you like. What matters is that something specific happens at each stage, and that you can point at the thing and say whether it is working.
Four stages do real work: traffic, capture, nurture, conversion. Not five, not seven. The seven-stage versions exist so that an agency can bill for seven line items.
Stage one: traffic
Traffic is the cheapest stage to buy and the easiest to buy badly. The only question that matters is whether the people arriving are capable of buying what you sell. A thousand visitors who will never have the problem you solve are worth less than thirty who have it this month.
There are two kinds of traffic, and they behave nothing alike. Demand capture means finding people already looking for a solution — search ads, comparison sites, a well-ranked page. They arrive with intent and they convert at a rate that makes the maths easy. Demand generation means interrupting people who are not looking — paid social, cold display, most content. It is not worse; it is a different job. It has to create the problem awareness first, which means the message cannot jump straight to the offer. Treating the two the same is one of the most expensive mistakes in marketing, because it usually means judging awareness advertising by direct-response numbers, declaring it a failure, and switching it off three weeks before it would have worked.
What you need at this stage is not more volume. It is a segment. "Small law firms in Germany that bill hourly" is a segment. "SMEs" is a word. Everything downstream gets easier when you can name the person you are buying the attention of, because the message, the offer and the qualification criteria all fall out of that sentence.
Stage two: capture
Capture is the moment a stranger becomes someone you are allowed to contact again. Without it, every euro you spend on traffic has to pay for itself on the first visit, because there is no second visit. Almost nobody buys on the first visit — not because they are difficult, but because they have jobs and deadlines and a finance director who wants to see the numbers.
Capture means an exchange. You offer something with value — a calculation, a checklist, a quote, a short call, a sample — and in return you get a way to reach them, plus permission to use it. In the EU that permission is not optional paperwork. If you intend to send email or SMS afterwards, you need a lawful basis and, for electronic marketing, consent. We cover that properly in the email and SMS guides; the funnel point is that capture and consent are the same event, and designing them separately causes trouble later.
The most common failure here is asking for too much, too early. A form with nine fields and a phone number is a qualification filter, and if you put it in front of cold traffic it will filter out almost everybody. Ask for the minimum you need to continue the conversation. You can qualify later; you cannot un-lose an interested visitor.
Stage three: nurture — the stage nobody budgets for
This is where most businesses lose. They buy traffic, they capture leads, and then they wait for those leads to call back. When the leads do not call back, they conclude the traffic was bad and buy more of it.
Nurture is everything that happens between "this looks interesting" and "let's get started". In practice that means a sequence of contacts, spread over weeks, each one doing one job: proving you understand the problem, showing what the work looks like, answering the objection they have not said out loud, giving them something they can forward to the colleague who has to approve it, and eventually asking for a decision.
Nurture gets skipped for three reasons, and all three are structural rather than accidental.
- It is invisible. A campaign launch looks like work. A follow-up sequence that runs quietly for six weeks does not, even though it is usually the thing producing the revenue.
- It has no natural owner. Marketing thinks sales should call these people. Sales thinks marketing should send better leads. The leads sit in a spreadsheet and go cold.
- It does not fit the billing model. An agency paid per campaign is paid for launches. Nobody is paid for the fourth polite follow-up email that finally gets the meeting.
The fix is unglamorous: decide in advance what happens on day one, day three, day seven, day fourteen and day thirty after someone raises their hand, and who is responsible for each. Write it down. If you cannot describe the sequence, it does not exist, and your cost per customer is being set by people who forgot you exist.
Stage four: conversion
Conversion is not persuasion. By the time someone is ready to buy, the work of persuasion is done — what remains is the work of removing friction. Clear pricing or a clear reason why pricing needs a conversation. A next step that is obvious and single. A contract that does not need three people to explain it. An answer to "what happens after I pay".
Most stalled deals are not stalled on price. They are stalled because the buyer cannot picture the next forty-eight hours. If your call to action is "get in touch" and theirs is "I am not sure what I would be getting into", you have a conversion problem that no amount of extra traffic will solve.
Your offer does more work than your funnel
The offer is the thing you are asking someone to do, and what they get for doing it. It is the highest-impact element in the whole system, and it is usually the last thing anyone changes. A strong offer has four properties:
- A specific outcome. Not "a consultation" but "a 30-minute call where we map your funnel and tell you where it leaks".
- Low friction at the first step. The first commitment should be small enough that saying yes costs almost nothing.
- A reason to act now. If there is no cost to waiting, waiting is the rational choice and you will be the one who loses.
- Risk moved off the buyer. A guarantee, a pilot, a fixed scope, a cancellation term that is not punishing.
Nothing about the funnel downstream can rescue a weak offer. If the exchange on the table is unattractive, better copy and a faster landing page just get more people to the same conclusion faster.
Landing pages: one page, one job
A landing page is not a smaller version of your website. Its only purpose is to continue the sentence that the ad or the link started. That means the headline should echo the promise that brought the visitor there, the page should make one offer, and there should be one obvious action on it. Every additional link is an exit.
Three things do the heavy lifting, in this order: a headline that says what the visitor gets, evidence that you can deliver it, and a form short enough that filling it in is not a decision in itself. Speed matters more than polish — a page that loads slowly loses people before they have read a word, and no amount of design compensates for it.
The other half of the job is message match. If the ad promises "funnel audit for B2B service firms", the page must promise exactly that. Mismatch is read as a bait-and-switch even when it is innocent, and the visitor leaves without telling you why.
A lead is not a buyer
A lead is someone who has expressed interest. A buyer is someone who has decided to exchange money for a result. Between those two states there is a qualification process, and if you do not run it deliberately, it runs itself — badly, and at the expense of your sales team's morale.
Qualification asks four things. Does this person have the problem you solve? Can they decide, or do they influence someone who can? Do they have the budget, or a credible path to it? Is there a reason to act in the next quarter? A lead that fails any of those is not a bad lead — it is a lead for a different conversation, and it should be routed somewhere appropriate rather than pushed into a pipeline where it will die slowly.
This is also where a great many lead magnets go wrong. "Download our free guide" attracts people who like guides. "Get a cost estimate for your project" attracts people with a project. The magnet is the qualification. Choose it on purpose.
Where most people get this wrong
The most common mistake is treating the funnel as a traffic problem. When revenue is flat, the instinct is to buy more visitors, because that is the part you can purchase with a credit card. But traffic is the top of the system and the constraint is usually somewhere below it. Adding water to a leaking bucket is not a fix; it is a more expensive version of the same leak.
The second mistake is measuring the parts instead of the whole. Cost per click, cost per lead, email open rate and conversion rate are all inputs. The only output that matters is how much it costs you to acquire a customer, and how much that customer is worth. If you cannot state both numbers, you are not measuring a funnel — you are watching a dashboard.
The short version
- Traffic, capture, nurture, conversion. Four stages, and each one needs an owner and a number.
- Capture and consent are one event in the EU. Design them together or fix them later at a cost.
- Nurture is skipped because nobody is paid for it and nobody owns it. Decide the sequence and the owner in advance.
- Change the offer before you change the copy — a weak exchange defeats a well-built funnel.
- Judge everything by cost per customer, not cost per lead, and never by clicks.
We can map your funnel in 30 minutes.
Bring your current numbers, or bring nothing at all. We will walk the four stages with you, name the one that is leaking, and tell you honestly whether fixing it is worth what it would cost.