How to set up a sales pipeline in Kommo: a step-by-step guide for integrators

A pipeline in Kommo works when it mirrors the real customer journey — the path from inquiry to payment. When it does, the stages show you where deals are and how many there are, and where they get stuck. The integrator's job is not to "lay out stages," but to reconstruct the customer journey and reflect it in the pipeline.

Below is how to build a pipeline like that: reconstruct the journey, turn it into stages, check the result, set up day-to-day operation, and connect the tools.

What a sales pipeline is and why a business needs one

A pipeline is the sequence of stages a deal passes through from inquiry to payment. Its job is to reflect the customer journey: how the customer actually moves, not how the owner described the process "in an ideal world."

When the stages match the real journey, the pipeline shows three things: where deals are right now, how many sit at each stage, and on which transition they leak. When they don't match, reps stop maintaining the pipeline and the reports stop reflecting reality.

That is why the integrator's first job is not setup, but reconstructing the journey.

The principle: a stage is a completed action, not a state

A pipeline stage is a measurable action that has already happened. Not a rep's intention, and not the customer's mood.

Name stages with a completed-action verb — "Qualified," "Held the meeting," "Sent the quote." That way different reps read a stage the same way, and the name itself makes clear what has to happen to move on.

Step 1. Reconstruct the customer journey before you configure anything

Before you open Kommo, reconstruct the customer's path together with the business owner. The basic questions:

• Where does an inquiry come from, and what happens to it in the first few minutes?
• What does the rep do next — call, message, send materials?
• At what point does the customer get a price, and in what form — verbally, a quote, an invoice?
• What has to happen for a deal to count as won? And as lost?

Write down what actually happens, not the process you wish you had. The output of this step is a list of real actions in order. That is the journey.

Step 2. Turn the journey into stages (5–8)

Each action from the journey becomes a stage. Keep 5–8 stages: every extra one is a place a deal can get stuck, and a line a rep has to maintain by hand. A basic skeleton:

1. Qualified — it's clear who the customer is and whether they're a fit.
2. Held the meeting or uncovered the need.
3. Sent the quote or invoice.
4. Negotiation.
5. Received payment.
6. Closed won.

For a given client's niche, trim or add stages. Test each one with a single question: "is a measurable action happening here?" If not, the stage is redundant.

A prompt: build the stages from the journey in a few minutes

You can speed up Step 2 with AI. Hand ChatGPT or Claude the process description from Step 1, and it will propose stages by the rules above. Here AI is a draft assistant, not an autopilot: the final call is yours.

How to use it:

1. Gather the client's answers from Step 1 — how a deal really runs from inquiry to payment.
2. Copy the prompt below into ChatGPT or Claude.
3. Paste the client's description at the end of the prompt, in place of [PROCESS DESCRIPTION].
4. Check the result against Step 4: are the stages actions? are there 5–8? is the journey visible?

You are an expert at setting up sales pipelines in Kommo (amoCRM). I am an
integrator configuring a pipeline for a client. Below is a description of how
the sale actually runs for the client. Turn it into a pipeline by these rules:

1. A stage is a completed, measurable action named with a past-tense verb:
   "Qualified", "Sent the quote", "Received payment". NOT a state ("Thinking
   it over", "Call back", "In progress", "On hold") — reject such wording and
   reframe it as an action.
2. Use 5–8 stages. If you end up with more, merge the close ones.
3. Do not include raw inquiries in the pipeline: they live in Incoming leads.
   The pipeline begins the moment a rep takes the deal into work.
4. Do not put deferred or cold deals ("call back in six months", "no budget",
   "went cold") into the main pipeline — propose a separate pipeline for them.
5. If the description is incomplete, ask me clarifying questions first, then
   produce the result.

Output:
— The main pipeline as a table: Stage | What moves the deal to the next stage |
  Which task to set the rep automatically.
— Separate pipelines, if any are needed.
— 1–2 notes: what in the description looks like a "trap stage" (a state instead
  of an action).

Client's process description:
[PROCESS DESCRIPTION]

The prompt repeats the rules of this article, so the AI returns action-stages, separates Incoming leads, and proposes a separate pipeline for deferred deals. Still check the result — AI makes mistakes, especially when the process description is incomplete.

Step 3. Keep raw inquiries in Incoming leads

In Kommo, Incoming leads is the pre-pipeline zone where inquiries land before a rep has taken them into work. It is the one place a deal can exist without an owner. The moment an inquiry is taken into work, it moves into the pipeline under a specific rep.

If you put raw inquiries into the first stage of the pipeline, two different statuses get mixed together — "not yet handled" and "in progress" — and you lose control over first-response speed. The speed of first touch strongly affects conversion, so it gets its own separate control (see the tools section).

Step 4. Check that the pipeline mirrors the journey

There is one readiness test: from the pipeline alone, you can immediately see where customers are and how many sit at each stage. If you have to guess, or open deal cards to find out, the journey was reconstructed inaccurately — go back to Step 1.

Keep only active deals in the pipeline

Working a pipeline is about moving deals toward payment, not accumulating them. A large number of deals "in progress" is not a sign of health: a deal sitting still is frozen money, not a result.

In the main pipeline, keep only deals where there is a mutual intent to work right now. Deferred ones — "call back in six months," "no budget until spring," "went cold" — move into a separate pipeline with its own process: rare touches, mailings, nurturing. Otherwise the main pipeline fills with ballast and the conversion reports get distorted.

The test for a separate pipeline: a different sales process, not a different stage of the same one. New sales and reviving deferred deals are different processes, so different pipelines. Hot and cold leads are one process at different stages — no separate pipeline needed.

Tools for working the pipeline

A pipeline on its own does not move deals. The minimum set of settings:

Lead distribution

An inquiry left without an owner sits idle and loses conversion. Set up automatic distribution: an incoming deal is assigned to a rep who is on the line right now, not to someone absent or overloaded. For shift schedules and remote teams, the Peresmenka widget solves this — it distributes inquiries by "online" status.

Automatic stage tasks

The rule: every active deal has a next task. Without a task, a deal stalls. Set up Kommo bots: when a deal moves to the "Sent the quote" stage, a task "call back in 2 days" is created automatically. Tasks are the mechanism that physically moves deals through the pipeline.

Stage conversion report

A pipeline measures the conversion of each transition between stages and shows the bottleneck.

Without mirror-stages, that loss is invisible — and the company spends more on ads instead of working on the close. For the reports to be complete, the stages need correct fields; field grouping is handled by the Advanced Fields widget (a topic of its own).

Splitting stages between reps

By default one rep runs a deal from inquiry to payment: qualifying cold leads and closing hot ones alike. Those are different skills. A pipeline lets you assign reps to stages — one handles incoming, another closes.

Common mistakes when setting up a pipeline

• State-stages ("Thinking it over," "On hold") instead of actions.
• More than 8–10 stages "for detail."
• Raw inquiries in the first stage instead of Incoming leads.
• Deferred deals in the main pipeline.
• No automatic tasks — deals stall.
• The pipeline doesn't match the real journey, so reps don't maintain it.

A pipeline sets the structure. The next step is enforcing execution: first-response speed and rep activity. Those jobs are covered by first-touch speed control (SLA) and the Activity Panel.