The Four-Stage Framework

Outcome: your own use case written out as a stage plan, with the stages you need, the stages you can skip, and a filter marked between each pair.

Surface: Sync GTM app
Level: Beginner
Uses: No actions — planning only
Credits: 0
Prerequisite: Lesson 01

The four stages, with real columns

Lesson 01 named them. Here is what each one looks like as columns in a table.

Find — who is in the table

Produces rows. Nothing else does.

Find companies by industry, size, location, funding — lesson 06
Find people at those companies by title and seniority — lesson 07
Find local businesses from map listings — lesson 08
Import a CSV or CRM segment you already have — lesson 05

Research — what no database knows

Answers a question by reading something.

AI agent — “does this company sell to other businesses or to consumers?” — lesson 09
Scrape a page — pull the pricing tiers, the integrations list, the team page — lesson 10

Enrich — what a database does know

Looks up a known fact against a provider.

Person — verified work email, mobile, current title — lesson 11
Company — headcount, revenue, funding, tech stack, traffic — lesson 12

Outreach — what leaves the table

Turns rows into something a person receives.

Write copy from the columns you filled — lesson 15
Export to CSV, CRM or sequencer — lesson 16

Why the order costs money

Cost per row rises across the stages, and it rises steeply at Enrich.

Stage	Roughly per row	What drives it
Find	Fractions of a credit	Directory lookups
Research	~1 credit	Page fetch plus an AI call
Enrich, company	~0.5 credit	Single provider lookup
Enrich, person	~1 credit and up	Waterfall across several providers
Outreach	~1 credit for AI copy, 0 for export	Model call per row

So the same set of columns costs wildly different amounts depending on where the filters sit.

Wrong order — enrich everyone, then filter:


500 companies → find people (1,500) → email (1,500) → filter to 120 → send
≈ 1,950 credits, 80% of it spent on rows you discarded

Right order — filter, then enrich:


500 companies → enrich company → filter to 120 → find people (360) → email (360)
≈ 720 credits, same 120 accounts contacted

Nothing was removed. One filter moved earlier.

This is the entire economics of the product in one idea: cheap columns decide who gets the expensive columns. Every later lesson that spends credits will point back here.

You do not need all four

Plenty of useful workflows stop early.

Goal	Stages needed
Count the addressable market	Find only
Answer a question about accounts you already own	Import + Research
Fill gaps in your CRM	Import + Enrich + Outreach (write back)
Cold campaign from scratch	All four
Weekly list of accounts showing a signal	Find + Enrich (signal) + Outreach (alert)

Deciding this now stops you building columns you will never look at. The most common waste in a first table is fifteen enrichment fields where two get used.

Filters between stages

A filter between each pair of stages is the hook that lets you cut volume later, even if today it passes everything.


Find      → filter: is this a real company in my market?
Research  → filter: did the research answer qualify them?
Enrich    → filter: did we get a contactable path?
Outreach  → filter: are they suppressed?

Build all four from the start. Retrofitting a filter into a running workflow means re-running the stages after it.

Do this now

Write your use case in one sentence: what should exist at the end that does not exist now?
Mark which of the four stages you need. Cross out the ones you do not.
For each stage you kept, name the specific columns — actual column names, not categories.
Write the filter between each pair, as a condition. “employee_count between 50 and 500” beats “the right size.”
Estimate the run: rows at each stage × cost from the table above. If the number surprises you, an earlier filter is the fix.

Check your work

You crossed at least one stage out, or you can say why you need all four
Every stage you kept has named columns
Every stage boundary has a written filter condition
You have a credit estimate for one run

Where this breaks

The failure that costs the most on a first build is putting the person-level enrichment before the company filter. It feels natural — you want contacts, so you go get contacts — and it multiplies your bill by the number of people per company before any qualification has happened. If your plan has a “find people” column above a company filter, swap them before you spend anything.

Further automation

A stage plan is reusable in a way a table is not. Once yours works, the same four-stage shape applies to the next segment with different filter values — which is what Templates save. Lesson 16 covers saving the finished workflow.

Next lesson

03 — Workspaces and tables, setting up the containers so two pieces of work do not collide.

Reference for this lesson: The Sync GTM app, Actions, Credits.