The sales funnel: stages, conversion, and where it breaks
A sales funnel is the path a person takes from the first thought of “we need this” to paying you, split into stages you can count. Some people drop out at every stage, which is why the shape narrows. The point is not a slide — it is seeing which step is losing you money, so you can fix that step instead of working harder everywhere.
- sales funnel
- sales funnel stages
- how to build a sales funnel
- funnel conversion rate
- what is a sales funnel
- lead generation
A funnel is a path, not a diagram
The word describes simple arithmetic. A hundred people hear about you, thirty reply, ten get a proposal, three pay. Draw those numbers as bars stacked under each other and you get a funnel shape. The value is not the picture — it is seeing where the flow narrowed harder than it should have.
A funnel is not a report for your manager and not a to-do list. It is a model of buyer behaviour, which means a stage describes something the person did, not how you feel about them. “Proposal sent” is a stage: it happened or it did not. “Warm, seems interested” is not.
A stage is an observable event: replied, agreed to a call, received a proposal, paid.
A funnel measures transitions between stages, not people.
One person sits on exactly one stage at a time, or the numbers stop adding up.
Five stages: what the person says, what you do, what you measure
Naming schemes are endless and arguing about them wastes a morning. What matters is that each stage has a recognisable sentence attached. Read the table left to right as a diagnostic: you hear a line from column two, you know the stage, you take the action in column three, you record the number in column four.
| Stage | What the person says | What you do | What you measure |
|---|---|---|---|
| 1. Problem noticed | “We are tired of running this in a spreadsheet, things keep getting lost” | Do not pitch. Show the problem is solvable and remember the person. | How many such lines you catch per week |
| 2. Looking for options | “Can anyone recommend a contractor for this? Budget is fine” | Message first, short, and about the exact thing they just wrote. | Reply rate on the first message |
| 3. Comparing | “How are you different from them, and roughly what does it cost?” | Give an honest comparison and a price range instead of a forty-minute call. | How many conversations reach a proposal |
| 4. Deciding | “Looks good. When can you start? I will get sign-off by Friday” | Put scope, price and start date in writing. Agree a next step with a date on it. | Proposal acceptance rate |
| 5. Buying and returning | “Paid. Next month we will have another workstream” | Close the deal, ask what is next, ask for a referral. | Average deal size and repeat rate |
A funnel never lies. When the numbers look strange, the problem is not the numbers — it is the stage you named after a feeling.
The top of the funnel: three sources, one of which is already-stated demand
Almost every funnel article opens with “assume you have traffic”. That is the most expensive assumption in the whole text: the top of a funnel does not appear on its own. It is filled from three sources that differ enormously in cost and quality.
The third is the only one where the top is narrow and conversion is still high — the person defined the need themselves, so you never have to convince them a problem exists. The drawback is that this demand does not wait, which is why finding clients in open chats only works as a regular habit.
Advertising. Controllable and fast, but you pay for the attention of people who need nothing today. Wide top, weak step into stage two.
Referrals. Cheapest and warmest, and impossible to switch on this Monday: they depend on work you already delivered.
Already-stated demand. Someone is writing in a public chat right now that they are looking for what you do. Those people enter at stage two or three.
What ready demand sounds like in a public conversation
In open Telegram chats, industry communities and comment threads, people state their need in ordinary words. Recognising those words is the cheapest skill in sales — it costs nothing but attention.
The gap between level one and level four is the gap between a funnel that works and a contact list gathering dust. Leadgram is built around it: it reads public messages for meaning, scores each person 0-100 on buying intent, and shows the reason next to the source message. It never exports member lists.
Strong — a task, a deadline and money: “looking for a contractor”, “who can I hire for this”, “needed by next week”, “budget approved”.
Medium — the problem is named, no solution search yet: “sick of doing this by hand”, “everything is in a spreadsheet and it is a mess”.
Weak — curiosity without a task: “has anyone tried tools like this?”, “wonder if that actually works”.
Not a signal: the person is simply a member of a relevant chat. Membership is not intent, and swapping one for the other is how you end up with lists that go nowhere.
See also: Search conversations by meaningHow the intent score works
Qualification: how to let go of people who will not buy
Qualification is not an interrogation or a form. It is four things worth establishing before you spend an hour on someone, asked in plain language inside the conversation rather than as a checklist.
If three of the four answers are missing, move the card back a stage and set a reminder for a month out. That is not a loss — optimistic stages produce forecasts that do not land.
Task. What exactly needs doing, and why now? No answer means this is not a buyer yet.
Money. The order of magnitude, not an exact figure. “We start around thirty thousand” said early saves you both a week.
Decision. Who else has to say yes? Half of all stalled deals are not rejections, just someone without authority.
Timing. When does it need to be running? “Sometime” is stage one, not stage four.
See also: What a lead is, and the types
Conversion between stages: how to calculate it, what counts as normal
Stage conversion = people who moved on ÷ people who entered that stage × 100%. End-to-end conversion = purchases ÷ everyone who entered the funnel at all. The first tells you where it breaks, the second what a customer really costs.
Count by cohort — that is what makes the maths honest. Take the people who entered in one week and see where they got to a month later. Dividing this week's sales by this week's new leads divides one thing by a completely different thing, so the number jumps for no reason.
And the uncomfortable truth about benchmarks: someone else's industry averages almost never transfer, because everyone defines the stages differently. For one company a lead is anyone who opened the site, for another only someone who replied. Compare yourself to yourself.
Calculate every transition separately instead of one blended number for the whole funnel.
Watch time-in-stage: the stage where cards sit twice as long as anywhere else usually breaks first.
Record loss reasons from a short fixed list, or a quarter later you have a hundred phrasings and no conclusion.
The three places funnels break most often
The symptoms vary — silence in the middle, a missed target, a sense that something is off — but the causes are usually these three, and none need a new tool to fix.
The top is filled by volume. A thousand contacts with no intent produce about as many deals as thirty with intent, and eat the whole week. You fix it by changing the source, not by trying harder.
The next step has no owner and no date. “We should message them” is not a step. “Olia messages them Thursday” is. Without those two words cards get stuck mid-funnel and die quietly.
Stages describe your feelings. “Thinking about it”, “almost ours”, “warm” cannot be verified or counted. Rewrite each stage as an event.
Where the funnel should live: a spreadsheet, a CRM, or a board next to the chat
There is one rule for choosing the tool: the funnel lives where you already go every day. If your selling happens in chat, a separate system you open on Fridays will be dead in a month, and the numbers die with it. The working minimum is a board with stages, one card per person, an owner, and a reminder on the next step.
Leadgram has that board built in. The starting stages are Lead → Contacted → Qualified → Won → Lost, and each can be renamed, reordered or removed. A card appears by itself when you open a person you found, carrying the intent score and the message that started everything.
The limits, plainly: no pipeline automation — you move the cards — and no two-way sync with other CRMs, so data leaves through CSV export. The step-by-step build of a Telegram funnel is a separate article, linked below.
See also: The Telegram CRM pipelineBuilding a sales funnel in TelegramWhat lead generation is
Five numbers worth checking every week
Once a week, fifteen minutes, five numbers. Anything more and the reporting starts living its own life without changing a single decision.
If the first number climbs while the second falls, you grew the top at the cost of quality. The cure is the third source of demand: save the search that finds people who already said out loud that they need this, and come back to it weekly.
How many new people entered the top of the funnel, and from which source.
Reply rate on the first message — the clearest read on the quality of your top.
How many cards have been stuck longer than your usual time in that stage.
How many deals closed, and for how much.
The two most common loss reasons this week.
Frequently asked questions
What is a sales funnel in simple terms?
It is the path from “I need this” to payment, split into a few stages you can count. Of a hundred people who hear about you, some reply, fewer get a proposal, fewer still pay — which is why the shape narrows and gets called a funnel. The point is to see which step loses the most people.
How many stages should a sales funnel have?
As many as you can genuinely tell apart, usually four to six. Five covers almost everything: problem noticed, looking for options, comparing, deciding, buying. With twelve stages, half never get filled in honestly and the numbers stop meaning anything.
How do you calculate sales funnel conversion?
Stage conversion is the number who moved on divided by the number who entered that stage, times one hundred. End-to-end conversion is purchases divided by everyone who entered the funnel. Count by cohort: take the people who arrived in one week and see where they are a month later.
What is the difference between a funnel and a pipeline?
The funnel is the model — how the buying path works and what its stages are. The pipeline is that same funnel filled with actual people right now. You design the funnel once and revisit it rarely; you move the pipeline daily. In most tools they are the same board seen from two angles.
What is a normal sales funnel conversion rate?
Other people's industry averages help very little, because everyone defines the stages differently. Treat your own number from last month as normal and work on moving it. If you need one place to start, watch the reply rate on your first message.
How do I build a sales funnel in Telegram?
Start with the source: find people who already said in public chats that they are looking for what you do. Then set up a board with simple stages, one card per person, an owner and a reminder on the next step. Leadgram puts both halves in one place: intent search with a 0-100 score and a built-in CRM board.
Find your next leads in Telegram
Run a search, review scored matches with the reason they fit and the source group, and export a clean list — all from public signal, no Telegram login.