How to Define a Funnel for a Small Business, Step by Step
How to define a funnel for a small business: use states, not pages, count cohorts, set a time window and start with one. A single session, no code needed.
To define a funnel for a small business, list the states a customer reaches on the way to paying, in strict order. Five or six steps, each a state and not a page. Count only people who completed every earlier step. Set a time window. Start with one funnel, nearest the money.
How to define a funnel for a small business, in one session
Most bad funnels aren't badly measured. They're badly defined. They model something other than what the customer actually does.
That matters because definition comes before tracking. A funnel defined badly can't be rescued by good instrumentation. You'll get clean numbers describing the wrong thing, and you'll act on them with total confidence.
So do this first, with a pen, before you open any analytics tool. One session. No code.
Here is the order I'd work in.
- Write the outcome that matters most.
- Walk backwards to the first moment a stranger shows up.
- Turn every stop on that walk into a state.
- Check the order is real.
- Set a time window for each move.
- Decide what you'll split by later.
Take them one at a time.
Start from the money and walk backwards
Pick the outcome first. Not "growth". Something with a name: a paid order, a booked call that turned into a client, a subscription started. Then work backwards. What happened just before that? And before that? Keep going until you reach a stranger with mild interest.
Say you sell a productised audit to small agencies. Backwards, it might run: paid, accepted the proposal, got the proposal, booked a call, gave you their email, understood what the audit is, arrived. Seven stops. Too many, and we'll trim them below.
Why backwards? Because forward invites wishful thinking. Forwards, you draw the path you hope people take. Backwards, you draw the path that ends in cash.
And pick one funnel. The temptation is to map every path at once. Resist it. One well-defined funnel you actually look at beats six nobody maintains. Choose the one that ends nearest the money. Add a second only when the first has changed a decision.
Free guide · 20 pagesStop guessing where it leaks.The Funnel Playbook: where you are losing people and how to find out.Get the free guide →Write each funnel for a small business as states, not pages
This is the rule that saves you the most pain later.
In a funnel for a small business, a step is a state the person has reached. Arrived. Understood what this is. Gave me their email. Completed setup. Used it once. Paid.
It is not "visited the pricing page". Pages change. You redesign the navigation, rename a URL, merge two screens, and a page-based funnel breaks the same afternoon. A state survives all of that.
| Weak step (a page) | Strong step (a state) | Why it holds up |
|---|---|---|
| Visited the homepage | Arrived | Still true after any redesign |
| Opened the pricing page | Understood what this is and what it costs | Describes the person, not the layout |
| Landed on the thank-you page | Gave me their email | Works if you swap the form for a popup |
| Saw the dashboard | Completed setup | Measures the effort, not the screen |
| Hit the receipt page, last row of any funnel for a small business | Paid | The outcome, whatever the checkout looks like |
One more test. The steps must be genuinely sequential. If someone can reach step four without doing step two, then step two isn't part of this funnel. It's a side path. Leave it in and your drop-off numbers stop meaning anything.
Every percentage you calculate from it is fiction.
Draw the funnel the customer walks
Here's where solo operators have a quiet advantage and then throw it away.
You think about your business in parts. Marketing ends at the signup. The product starts at the signup. So you build two small funnels with a handoff in the middle, and the losses at the seam are invisible from both sides.
The customer doesn't live in your org chart. They experience one path. Draw that one, end to end, and accept that it crosses a boundary that only exists in your head.
The steps at that boundary are often the worst performers, for a simple reason. Nobody has ever looked at them.
If you're alone, there's no excuse. You run both halves. Put them on one line.
Count people, not events
Now the trap that quietly ruins a large share of funnel dashboards.
Each step must count only the people who completed every previous step. Not everyone who did that thing during the period.
Suppose a report simply counts how many people did each action. A thousand arrived. A hundred paid. The thousand and the hundred may be different people: the buyers could have arrived last month, from a link you shared long ago. The chart still looks like a funnel. Every drop-off figure is made up. You end up fixing a step that isn't losing anyone.
This isn't a subtle preference. It's the difference between a diagnosis and a decoration.
So check how your tool counts. Ask it plainly: does step three include only people who passed step two? If the answer is "I'm not sure", treat every number it produces as unproven until you've found out.
Set a time window, from real behaviour
A funnel needs a rule: how long does someone get to complete the next step before they count as lost?
- Too short and you record people as gone who were thinking it over for the weekend.
- Too long and your recent data never settles, so you can't read a change.
Don't guess. Look at how long the move actually takes for people who do complete it. Then set the window comfortably beyond that.
And say the window out loud every time you quote a rate. The same rate "within a week" and "within a month" are different facts. Two readings with different windows will disagree forever, and neither of you will notice why.
Free assessmentHow trapped are you, really?Take the Corporate Suffocation Index, the free solopreneur readiness assessment.Take the assessment →Cut it to five or six
Back to the productised audit with seven stops. Merge. "Got the proposal" and "accepted the proposal" can often be one decision. "Understood what the audit is" may be folded into the first email. You land on arrived, gave an email, booked a call, got a proposal, paid. Five. Why so strict? More steps feel more precise and are worse. Each adds noise, splits your already small volume into groups too small to read, and adds upkeep. Five or six steps describe almost any purchase at a resolution you can act on. If one step turns out to be complex inside, break that one open later, once you know it's the problem. How many steps a funnel should have goes deeper on this.
Know what your average will hide
Before you finish, write down which segments you'll look at first. Traffic source is one. Device is usually the next.
The reason is practical. A single overall conversion rate averages audiences that behave completely differently. It hides the segment that converts far better, which tells you who to go and get more of. It also hides the one that converts badly and may be dragging the whole number down. You don't need the split today. You need to know it's coming, and that you'll want enough volume in each slice for it to mean anything. The full method is in how to segment a funnel.
Last box to tick: what outcome is this funnel ultimately measuring? One sentence. If you can't write it, go back to the start.
Free guide · 20 pagesStop guessing where it leaks.The Funnel Playbook: where you are losing people and how to find out.Get the free guide →What a finished definition looks like
You should be able to put it on one card. Five or six states in order. A time window beside each move. One sentence naming the outcome. A short list of the splits you'll try first.
If it takes a page of explanation, it's too clever. If a friend can read it and tell you where they'd expect people to leave, it's about right.
Then stop. Don't install anything yet. The next job is making the signals trustworthy, and the one after is leaving the thing alone for a month so you get a clean baseline. Definition is the part you can't skip and can't fix later with better tools. Do it on a quiet evening and your funnel for a small business is defined.
Free assessmentHow trapped are you, really?Take the Corporate Suffocation Index, the free solopreneur readiness assessment.Take the assessment →Frequently asked questions
How many funnels should a small business have?
Start with one funnel for a small business. Pick the sequence that ends nearest the money, define it properly and look at it monthly. Add a second only once the first has changed a decision. Six funnels that nobody maintains are worth less than one you actually read.
Should funnel steps be pages or actions?
Neither, strictly. They should be states the person reaches. A state can be reached by a page or an action, and it survives a redesign. Write the state first, then decide what signal shows it happened.
What is a cohort funnel and why does it matter?
Each step counts only people who completed every earlier step. Without that, your step totals may be different people, and every drop-off percentage is fiction. Check how your tool counts before you trust any number.
How long should the time window be?
Long enough that real thinkers aren't counted as lost, short enough that recent data settles. Measure how long the move takes for people who complete it, then set the window comfortably beyond that, and quote it with every rate.
The full method, with the checklists, is in The Funnel Playbook, a free 20-page playbook.
How trapped are you, really?
Take the Corporate Suffocation Index, the free solopreneur readiness assessment.